Compare commits

..
Author SHA1 Message Date
veronica 98f3384554 docs(W4): mark the Applied table as first-pass and point it forward (fred's F1)
ci/woodpecker/pr/ci Pipeline was successful
The Applied table carried 107 stamped and 17 held, the counts from the first
pass, with nothing telling a reader they had moved. The re-closed arithmetic at
the end of the document was already correct, so the document was not wrong
overall, but a reader hitting the table got a stale count with no forward
pointer. In this PR specifically that is the defect the PR exists to remove,
appearing in the PR's own worklist.

The first-pass numbers are left standing and labelled rather than overwritten,
matching every other correction in this document: a record of what a pass
produced is worth more than a number silently updated to still look right. The
current census follows it, re-measured: 108 stamped + 16 + 3 + 1 = 128, kinds 54
guide / 34 record / 13 spec / 7 tracking, status 106 active / 2 completed, and
the two files carrying source_of_truth: true named.
2026-08-20 20:46:16 -05:00
veronica 4211fb87f8 docs(W4): withdraw the test count from the method paragraph, do not replace it
ci/woodpecker/pr/ci Pipeline was canceled
The paragraph arguing 'run the suite before concluding a docs change has no code
consumers' carried a number: 'CI 2592 ran 1617 tests, 1 failed, 1616 passed'.
That number is wrong and is withdrawn in place rather than deleted.

Extracting per-test totals from these pipeline logs is not a reliable method:
the same regex over the same log format returns 1003 for 2592, 1022 for 2593
and 3471 for 2594, which are runs of the same suite. Three irreconcilable
answers from one method is proof the method does not measure what it claims,
so no corrected count is substituted.

The claim is restated in what the log carries reliably, the named FAIL list and
the turbo task line. The count was decoration on a claim already carried by a
specific name: one named failing spec is what refuted three hand-searches.
2026-08-20 20:41:52 -05:00
veronica ff40794bfe docs(W4): revert channel-protocol to guide (F withdrawn), stamp docs/README.md as the contract's source of truth (Q1)
ci/woodpecker/pr/ci Pipeline was successful
fred withdrew decision F after re-measuring channel-protocol.md with a control:
zero uppercase RFC2119 terms, seven lowercase 'must' all disclaiming authority,
under a banner refusing requirements status. F rested on 'the doc graph outranks
the page's own banner', which contradicts decision D, 'kind follows content'.
D is the rule; F was its counterexample. Neither rescue (vision's adjective edit
on the citing docs, or a banner edit) was taken: a kind that survives only by
editing the evidence around it is not a classification.

Q1 answered: docs/README.md prescribes the contract, so by D its content is a
spec, and by B it is the source of truth for that contract. A plan never
outranks the artifact it planned, so the flatten plan's stale 7-value enum is
corrected there rather than competing here.

128 live docs = 108 stamped + 16 operator-held + 3 supersede deferrals + 1 generated.
2026-08-20 20:21:06 -05:00
veronica 94a7d5b692 test(mosaic): pin quickstart.md to status: active, the contract's value for in-force
ci/woodpecker/pr/ci Pipeline failed
CI 2592 proved the old front-matter schema HAD a consumer. vision's C2 concluded "no consumer
found", fred's decision A rested on "no installed base to protect", and I accepted both. All three
of us were wrong.

installation-documentation.spec.ts:39 asserted a raw regex over the markdown text:

    expect(markdown).toMatch(/^---\n[\s\S]*?\nstatus: current\n[\s\S]*?\n---\n/);

Not a front-matter parse and not a key lookup, which is why a search organised around parsing
primitives could not see it. Stamping docs/USER-GUIDE/getting-started/quickstart.md replaced
`status: current` with `status: active` and turned it red.

Updated to the contract's vocabulary, intent unchanged: the page must still declare itself in
force, and must still teach no curl-pipe-to-shell install. Verified by evaluating both regexes
against the real file: old false, new true. Control: a page carrying `superseded-by` still fails
the new regex, so the assertion still asserts something.

2592 also confirms the NORTH_STAR revert took: the north-star drift test is green, `format` and
`lint` are green, and this was the only failure in 1617 tests.

Method point worth keeping: the full suite is a stronger consumer search than any grep, because it
does not depend on guessing how a consumer reads the file.
2026-08-20 20:04:35 -05:00
veronica 12d5258e20 docs(W4): apply fred's six contract decisions from PR #1350 comment 23693
A - docs/README.md:149-190 rewritten. It prescribed a competing front-matter schema
    (title/type/audience/status/source_of_truth) adopted by 4 of 128 live documents. Two
    documented conventions in one repo is the defect this pass removes, so the README now
    documents the contract and the 4 files convert in the same commit: `type` dropped
    (kind replaces it), `title`/`audience`/`source_of_truth` kept.
B - source-of-truth leaves the kind enum, which is now 6 values, and returns as an orthogonal
    boolean. kind was carrying two independent facts. docs/requirements/native-kanban-sot.md
    is stamped `kind: spec` + `source_of_truth: true`, which is what it always was.
C - status gains `completed`. Applied to the two executed plans, on artifact evidence rather
    than on their own say-so: --purpose push|merge ships in ci-queue-wait.sh, and every section
    the README plan specifies exists in docs/README.md today.
D - kind follows content, never filename. docs/native-kanban-sot/TASKS.md is `kind: spec`
    because its body says "a build plan, not a task tracker". The name stays wrong; that is a
    rename and it is out of scope here.
E - the contract covers .md only, written into the README as a decision with vision's
    YAML.parse measurement as the reason, so the omission does not read as an oversight.
F - channel-protocol.md guide -> spec. Applied, with a correction the reviewer should see: the
    ruling cites "7 normative MUSTs" and there are ZERO uppercase RFC2119 terms in that file.
    Control: the identical grep returns 25 lines in docs/requirements/native-kanban-sot.md. The
    citation half of the finding does hold and is larger than stated. Consequence recorded in
    the worklist: the file's own banner now contradicts its header.

Verified: 128 live .md under docs/ (127 baseline + this PR's worklist), 107 stamped, 0 invalid
kinds, 17 operator-held + 3 supersede-stamp deferrals + 1 generated = 21 unstamped. 107+21=128.
Control: the verifier reports valid=False when a kind is corrupted to `nonsense`, so the
0-invalid result is a real result. prettier --check clean across docs/.
2026-08-20 19:58:17 -05:00
veronica f6fbeaf57a docs(W4): response to the W5 adversarial pass, in the diff rather than a PR comment
pr-review.sh -n 1350 -a comment fails with HTTP 401 'invalid username, password or token'.
Measured cause with a control: veronica's seat token returns 200 on /api/v1/user and the shared
broker credential gitea-mosaicstack returns 401 on the same endpoint. The wrapper logic is fine,
the credential it is handed is dead, so the PR wrappers are down for every seat on this host.
MOSAIC_GIT_IDENTITY appears nowhere in credentials.sh, which is FD-9.

Constitution gate 8 makes that operation blocked, so the corrections land in the branch instead,
where they survive the PR anyway.
2026-08-20 19:55:23 -05:00
veronica 8a55c04108 docs(W4): prettier the worklist table padding
ci/woodpecker/pr/ci Pipeline failed
CI 2589 step `format` failed on exactly one file: this one, the plan document I hand-wrote.
prettier --check reproduces locally, rc=1, same single file.

That is worth stating for the review: the other 103 stamped documents pass
`prettier --check` unchanged, so the `---\nkind:\nstatus:\n---` front matter block is
already prettier-clean. The formatting failure was in my prose, not in the contract header.

Diff is 31 lines in, 31 out, all table column padding. No content change.
2026-08-20 19:48:07 -05:00
veronica bea47543f3 docs(W4): revert the NORTH_STAR.md stamp; a projection cannot carry a hand-added header
ci/woodpecker/pr/ci Pipeline was canceled
Found by vision's adversarial pass (W5) and re-measured here. The finding was raised as
latent; it is live in this PR.

docs/fleet/NORTH_STAR.md is generated by renderNorthStarMarkdown()
(packages/mosaic/src/commands/fleet.ts:373), whose first emitted line is the H1 and which
emits no front matter at all. fleet-north-star.spec.ts:110-114 is a drift test that reads
the committed file from disk and asserts full-string equality against the renderer output:

    expect(rendered).toBe(committed);

Stamping the file changed line 1 from '# Mosaic Fleet - NORTH STAR' to '---', so the
assertion fails. vitest could not be run in this worktree (node_modules empty, 'Cannot find
package yaml'), which is a harness gap and not a test result, so this is shown by
construction: the assertion above, the renderer's first lines.push, and line 1 before
versus after.

Reverted to origin/next verbatim: git diff origin/next -- docs/fleet/NORTH_STAR.md is 0
lines. Control on docs/fleet/reference/cli.md returns 13 lines, so the diff command does
report differences when they exist.

The header for a projection has to come from the renderer, not the file. That is a code
change and belongs in the flatten PR beside the resolveNorthStarPaths() fix.

Stamped count 104 -> 103. Untouched 23 -> 24. Worklist records it as a fourth held row.
2026-08-20 19:46:24 -05:00
veronica 37cd00e60d docs(W4): operator worklist for the 19 held rows, plus two contract conflicts
ci/woodpecker/pr/ci Pipeline failed
Records what the stamping pass applied, what it held and why, and the two
places where the contract contradicts itself or the repository:

- docs/native-kanban-sot/TASKS.md declares itself a build plan; the contract
  says TASKS.md is tracking and never a spec.
- docs/requirements/native-kanban-sot.md is hand-authored prose canon; plan
  section 5.2 calls that source-of-truth, the contract table requires
  source-of-truth to be machine-readable.
- docs/README.md already documents a competing front-matter convention on a
  colliding status key, applied to 4 files, with no code consumer.
- The contract has no defined form for a .yaml document, including
  docs/fleet/NORTH_STAR.yaml itself.
2026-08-20 19:31:25 -05:00
veronica f0d2dd9920 docs(W4): stamp kind and status front matter on 104 live documents
Applies the document contract from
docs/plans/2026-08-20_stack-docs-flatten-and-alignment.md section 3, partially:
`kind` and `status` only. `parent` is deliberately held until the flatten in
section 4 lands, so that 127 documents do not have to be re-pointed by hand
when docs/fleet/NORTH_STAR.yaml moves to docs/NORTH_STAR.yaml.

Scope, measured on origin/next at 63069149:

  127 live docs   = all *.md under docs/ minus docs/archive/ minus docs/_old_structure/
  104 stamped     here
   19 held        operator judgement (plan section 9), worklist in the same PR
    3 held        the SUPERSEDED TASKS.md stamps, which cite the moving path
    1 untouched   docs/fleet/FLEET-DOCTRINE.md, already stamped in W1

Kinds applied: 54 guide, 34 record, 9 spec, 6 tracking, 1 projection.
Every row carries a confidence and a one-line rationale in the worklist.

Two collisions with the existing state, both flagged rather than resolved:

1. docs/README.md:150-160 already documents a front-matter convention
   (title/type/audience/status/source_of_truth) with its own allowed values.
   It is applied to 4 of 127 files. Its `status` vocabulary is
   current|draft|deprecated|historical; the new contract's is active|superseded-by.
   The key collides. This commit lets the new contract win and rewrites
   `status: current` to `status: active` on those 4 files, keeping their other
   legacy keys untouched. No code reads any of them: `git grep source_of_truth`
   outside docs/ returns nothing. docs/README.md still prescribes the old
   convention and is an operator row, so it is not edited here.

2. Two of the plan's 20 operator rows are YAML files, not markdown
   (docs/fleet/examples/roster-v2.yaml, docs/openapi-tess.yaml), and the
   contract's front-matter form has no defined meaning for a .yaml document.
   That gap also applies to docs/fleet/NORTH_STAR.yaml, the source of truth
   itself. Raised in the worklist.

A third row from the plan, docs/fleet/north-star.md, no longer exists: W1
renamed it to docs/fleet/FLEET-DOCTRINE.md.

Verification: 104/104 parse with the expected kind and status in front matter;
the check was shown to reject a wrong kind before it was trusted. The diff
removes 4 lines total, all of them `status: current`.
2026-08-20 19:30:25 -05:00
47 changed files with 204 additions and 4248 deletions
+1 -8
View File
@@ -1,11 +1,4 @@
{
"schema_version": 2,
"integration_trunk": "next",
"release_branch": "main",
"flow": "trunk-release",
"canonical_remote": "https://git.mosaicstack.dev/mosaicstack/stack",
"canonical_clone": "host:/src/mosaic-stack",
"worktree_root": "host:/src/mosaic-stack-worktrees",
"worktree_policy": "orchestrator-precreated",
"notes": "next=development/integration; main=production release. Never branch work off main. worktree_policy is TRANSITIONAL: the wrapper worktree consumer is BLOCKED on the J3/#1174 amendment (checked roots + capacity guard); pre-creation is the interim orchestration choice, not closed policy — it becomes a timing choice only after the wrapper can validate this root."
"release_branch": "main"
}
-43
View File
@@ -91,15 +91,6 @@ steps:
# and sandboxes a throwaway git repo, so it resolves no real credentials and
# joins CI directly rather than the exclusions file.
- bash packages/mosaic/framework/tools/git/test-issue-close-fail-closed.sh
# Hermetic regression for the git identity ladder (#1356): mock tea on PATH,
# sandboxed repo, no real credentials (3/3 green under an empty HOME). Pins
# fail-closed: a seat whose login is missing gets a named error, never a
# borrowed identity. Joins CI directly; its #1007 exclusion is burned down.
- bash packages/mosaic/framework/tools/git/test-gitea-login-resolution.sh
# Hermetic regression for issue-view.sh (#1357): mock tea/curl, sandboxed
# repo. Pins that comment BODIES render on both paths and that a tea
# failure is named as what it was (git-config vs credential).
- bash packages/mosaic/framework/tools/git/test-issue-view-comments.sh
# Hermetic behavioural regression for the PreToolUse wrapper guard: proves
# it still blocks the three mistakes AND still lets reads, unwrapped
# endpoints and ordinary commands through. Both directions are asserted —
@@ -113,40 +104,6 @@ steps:
# stub supplies the scale instead of the host's own checkout.
- bash packages/mosaic/framework/tools/git/test-mosaic-worktree-large-repo.sh
# Canonical repo-structure declaration gate (T51 WP5c, spec §5.4 point 2):
# .mosaic/repo.json is the machine-readable structure SSOT consumed by git
# wrappers and the T32 gate seat; this is its repo-side CI enforcement.
# Path-conditional: runs when the declaration, the vendored validator, or this
# pipeline config changes (manual runs always include it). Fails the pipeline
# on any VALIDATION_ERROR and enforces the schema_version 2 authoring rule
# (--require-v2: edited/new declarations may not stay v1). The validator is
# vendored into the framework tree (spec §5.1 final home) — provenance in its
# header; the hostile-input suite (101 arms, hermetic) runs alongside so the
# gate's own instrument ships in the same commit as the gate.
structure-declaration:
image: *node_image
commands:
- apk add --no-cache bash git
# MOSAIC_HOST_ROOT is a runtime anchor (spec §1.2a: unset fails closed
# for managed validation). CI has no host, so the step provisions an
# EXPLICIT fixture root — honest configuration for the resolution path,
# never a guess about a real host; the per-host containment checks are
# runtime concerns and do not run against a fixture. Grammar, schema,
# refs, flow, remote normalization, and path grammar all prove here.
- mkdir -p /tmp/t51-ci-hostroot
- bash packages/mosaic/framework/tools/structure/validate-repo-json.sh .mosaic/repo.json --require-v2
- bash packages/mosaic/framework/tools/structure/test-validate-repo-json.sh
environment:
MOSAIC_HOST_ROOT: /tmp/t51-ci-hostroot
when:
- event: pull_request
path:
include:
- '.mosaic/repo.json'
- 'packages/mosaic/framework/tools/structure/**'
- '.woodpecker/ci.yml'
- event: manual
# Canonical verify:release stage `upgrade-guard`.
# Blocking gate (#791): a framework upgrade must never write or delete an
# operator-owned path. The HARD GATE proves an unanticipated operator sentinel
@@ -111,7 +111,7 @@ approve path carries the trap.) `pr-review.sh` sends the correct token for the d
Whatever you use, re-read `GET /pulls/{n}/reviews` and assert the state before reporting a verdict
placed.
The guard exits nonzero for any provider-asserted non-green, missing, or malformed CI state. If credentials or the provider are unavailable, it emits `CANNOT_ASSERT` and writes a JSONL audit record. Push degrades to exit 0 so recovery work is not bricked; merge holds with retryable exit 75 until the provider recovers, then self-clears without manual reset. Neither outcome is evidence that CI was clear. For a repository with no CI configured at all, `pr-merge.sh --no-ci-expected` is the sanctioned merge path: it forwards to `ci-queue-wait.sh --no-ci-expected`, which reclassifies a zero-context merge head as queue-clear only when the acting token holds repository admin and `MOSAIC_GIT_IDENTITY` names the asserting identity (a caller without one is refused with exit 78 before the admin lookup), and records the assertion (or its refusal) in the same JSONL audit log. `pr-merge.sh` automatically inspects the exact PR head repository and full commit SHA rather than its `main` base; this also handles fork PRs without branch-name ambiguity. Pass `--expect-head <approved-full-sha>` to bind a commit-specific review or merge-gate verdict; Gitea uses atomic `head_commit_id` and GitHub uses `--match-head-commit`.
The guard exits nonzero for any provider-asserted non-green, missing, or malformed CI state. If credentials or the provider are unavailable, it emits `CANNOT_ASSERT` and writes a JSONL audit record. Push degrades to exit 0 so recovery work is not bricked; merge holds with retryable exit 75 until the provider recovers, then self-clears without manual reset. Neither outcome is evidence that CI was clear. `pr-merge.sh` automatically inspects the exact PR head repository and full commit SHA rather than its `main` base; this also handles fork PRs without branch-name ambiguity. Pass `--expect-head <approved-full-sha>` to bind a commit-specific review or merge-gate verdict; Gitea uses atomic `head_commit_id` and GitHub uses `--match-head-commit`.
### Code Review (Codex)
@@ -219,23 +219,6 @@ Multi-instance support: `-a <instance>` selects a named instance (e.g. `personal
~/.config/mosaic/tools/health/stack-health.sh -f json
```
### Repo Structure Declaration (T51)
```bash
# Validate a .mosaic/repo.json declaration (schema v1/v2, host:/ grammar,
# ref grammar, cross-field rules, remote normalization; spec §5)
~/.config/mosaic/tools/structure/validate-repo-json.sh <repo>/.mosaic/repo.json
# CI authoring rule: new/edited declarations must be schema_version 2
~/.config/mosaic/tools/structure/validate-repo-json.sh <repo>/.mosaic/repo.json --require-v2
# Display mode (warns and omits root-dependent checks when MOSAIC_HOST_ROOT unset)
~/.config/mosaic/tools/structure/validate-repo-json.sh <repo>/.mosaic/repo.json --mode display
# Hermetic hostile-input suite (101 arms)
~/.config/mosaic/tools/structure/test-validate-repo-json.sh
```
### Shared Credential Loader
```bash
@@ -1,181 +0,0 @@
#!/usr/bin/env bash
# seat-logins.sh — project seat credentials into tea's login config.
#
# Issue: mosaicstack/stack#1356 (tea login resolution fails open).
#
# WHY THIS EXISTS. tea 0.14.0 has no --token on its operations; it can only use a
# login already stored in ~/.config/tea/config.yml. So the wrappers cannot read the
# seat secrets dir on the tea path. The secrets dir stays authoritative and this
# script projects it into tea's config, which is a DERIVED CACHE: regenerate it,
# never hand-edit it. Same shape as the config-registry projector, same reason —
# a third-party tool that cannot read our store has to be fed.
#
# Canonical login name is "<instance>-<seat>", which is what the identity ladder in
# detect-platform.sh computes from the seat name. A login the ladder cannot compute
# is a fail-open surface, so an ad-hoc name is a defect, not a style.
#
# COLLISIONS. tea refuses to store one token under two names ("token already been
# used, delete login 'X' first"). A hand-made alias holding a seat's token there-
# fore BLOCKS its canonical name. Detected up front by hashing, so a dry run shows
# it; --adopt resolves it by deleting the alias and re-minting canonically. Same
# token, same access, only the label changes.
#
# Tokens are never printed, never logged, and never passed on a visible command
# line beyond tea's own --token, which is unavoidable with this client. tea's
# stderr is echoed on failure with any token-shaped string redacted.
#
# Usage:
# seat-logins.sh # dry run, all seats (default: changes nothing)
# seat-logins.sh --apply # mint/refresh all seats
# seat-logins.sh --seat <seat> # limit to one seat
# seat-logins.sh --apply --adopt # also rename ad-hoc aliases to canonical names
set -euo pipefail
BRAIN_HOME="${MOSAIC_BRAIN_HOME:-$HOME/.mosaic}"
TEA_CONFIG="${TEA_CONFIG:-$HOME/.config/tea/config.yml}"
APPLY=0
ADOPT=0
ONLY_SEAT=""
# Instance -> server URL.
#
# Instances are named here because there is no registry to read them from yet.
# Override per-instance without editing this file, which is how a deployment adds
# its own hosts: MOSAIC_GITEA_URL_<INSTANCE>=https://...
declare -A INSTANCE_URL=(
[mosaicstack]="https://git.mosaicstack.dev"
[usc]="https://git.uscllc.com"
)
while [ $# -gt 0 ]; do
case "$1" in
--apply) APPLY=1; shift ;;
--adopt) ADOPT=1; shift ;;
--seat) ONLY_SEAT="${2:?--seat needs a name}"; shift 2 ;;
-h|--help) sed -n '2,33p' "$0"; exit 0 ;;
*) echo "seat-logins.sh: unknown argument '$1'" >&2; exit 2 ;;
esac
done
command -v tea >/dev/null || { echo "seat-logins.sh: tea not on PATH" >&2; exit 1; }
url_for() {
local inst="$1" ovr
ovr="MOSAIC_GITEA_URL_$(printf '%s' "$inst" | tr '[:lower:]-' '[:upper:]_')"
if [ -n "${!ovr:-}" ]; then printf '%s' "${!ovr}"; return 0; fi
printf '%s' "${INSTANCE_URL[$inst]:-}"
}
# Redact anything token-shaped before any tea output reaches a log.
redact() { sed -E 's/[A-Za-z0-9]{30,}/<REDACTED>/g'; }
# token sha256 -> login name, for every login tea already holds. This is what
# makes collisions visible in a DRY RUN instead of only as an apply-time error.
declare -A TOKEN_OWNER=()
if [ -r "$TEA_CONFIG" ]; then
while read -r sha lname; do
[ -n "${sha:-}" ] && TOKEN_OWNER["$sha"]="$lname"
done < <(python3 - "$TEA_CONFIG" <<'PY'
import sys, yaml, hashlib
try:
cfg = yaml.safe_load(open(sys.argv[1])) or {}
except Exception:
sys.exit(0)
for l in (cfg.get('logins') or []):
t = l.get('token')
if t:
print(hashlib.sha256(t.encode()).hexdigest(), l.get('name'))
PY
)
fi
minted=0; refreshed=0; skipped=0; failed=0; planned=0; adopted=0; blocked=0
existing="$(tea login list --output simple 2>/dev/null | awk '{print $1}' || true)"
shopt -s nullglob
for tokfile in "$BRAIN_HOME"/fleet/agents/*/secrets/gitea-*.token; do
seat="${tokfile#"$BRAIN_HOME"/fleet/agents/}"; seat="${seat%%/*}"
[ -n "$ONLY_SEAT" ] && [ "$seat" != "$ONLY_SEAT" ] && continue
base="$(basename "$tokfile" .token)" # gitea-<instance>-<seat>
inst="${base#gitea-}"; inst="${inst%-"$seat"}"
name="${inst}-${seat}"
url="$(url_for "$inst")"
if [ -z "$url" ]; then
echo " SKIP $name — no URL known for instance '$inst' (set MOSAIC_GITEA_URL_${inst^^})"
skipped=$((skipped+1)); continue
fi
if [ ! -r "$tokfile" ]; then
echo " SKIP $name — token not readable"
skipped=$((skipped+1)); continue
fi
action="mint"
grep -qx "$name" <<<"$existing" && action="refresh"
# Is this exact token already stored under some OTHER name?
tsha="$(sha256sum < "$tokfile" | awk '{print $1}')"
owner="${TOKEN_OWNER[$tsha]:-}"
collision=""
[ -n "$owner" ] && [ "$owner" != "$name" ] && collision="$owner"
if [ "$APPLY" -eq 0 ]; then
if [ -n "$collision" ]; then
if [ "$ADOPT" -eq 1 ]; then
echo " PLAN adopt $collision -> $name ($url)"
else
echo " BLOCK $name — token already stored as '$collision'; re-run with --adopt"
blocked=$((blocked+1)); continue
fi
else
echo " PLAN $action $name -> $url"
fi
planned=$((planned+1)); continue
fi
if [ -n "$collision" ]; then
if [ "$ADOPT" -eq 0 ]; then
echo " BLOCK $name — token already stored as '$collision'; re-run with --adopt"
blocked=$((blocked+1)); continue
fi
tea login delete "$collision" >/dev/null 2>&1 || true
action="adopt"
fi
# tea has no idempotent add; refresh is delete-then-add so a rotated token lands.
[ "$action" = refresh ] && tea login delete "$name" >/dev/null 2>&1 || true
if err="$(tea login add --name "$name" --url "$url" \
--token "$(cat "$tokfile")" --no-version-check 2>&1 >/dev/null)"; then
case "$action" in
mint) minted=$((minted+1)) ;;
refresh) refreshed=$((refreshed+1)) ;;
adopt) adopted=$((adopted+1)) ;;
esac
if [ "$action" = adopt ]; then
echo " OK adopt $collision -> $name ($url)"
else
echo " OK $action $name -> $url"
fi
else
# A failure here is real information: the seat's token is dead, or the server
# refused it. Do not paper over it; the seat cannot act until it is reminted.
# tea's own words, redacted — a summarised FAIL hides whether the cause is the
# credential or the client, which cost a diagnosis on 2026-08-21.
echo " FAIL $action $name -> $url"
echo " tea: $(printf '%s' "$err" | redact | head -1)"
failed=$((failed+1))
fi
done
echo
if [ "$APPLY" -eq 0 ]; then
echo "dry run: $planned login(s) would be written, $skipped skipped, $blocked blocked."
[ "$blocked" -gt 0 ] && echo "re-run with --adopt to rename ad-hoc aliases to canonical names."
echo "no changes made. re-run with --apply."
else
echo "minted=$minted adopted=$adopted refreshed=$refreshed skipped=$skipped blocked=$blocked failed=$failed"
fi
[ "$failed" -eq 0 ] && [ "$blocked" -eq 0 ]
@@ -257,36 +257,8 @@ assert_owned_tmux_server() {
fail "tmux server ownership or environment validation failed"
}
# Lease-broker socket preflight (#1292). The gated runtime (`mosaic yolo …` →
# launch-runtime.py) registers with the broker or dies ~4 seconds in, with the
# diagnostic invisible because tmux destroys the dead pane. This check runs
# BEFORE any tmux effect — including the ownership probe below — so a host
# without a broker produces a named, surviving refusal instead of a doomed
# pane. Exit 75 (EX_TEMPFAIL), distinct from 64 (bad projection) and 69 (host
# not ready for other reasons); the agent@ unit is Type=oneshot with no
# Restart=, so the failed unit keeps its message instead of looping. Socket
# resolution matches launch.ts's defaultLeaseBrokerSocket precedence exactly.
# This preflight DETECTS and REFUSES — it never starts the broker (activation
# belongs to the fleet control plane; a component that both detects and fixes
# cannot be used to measure whether the fix worked).
broker_socket_path() {
if [ -n "${MOSAIC_LEASE_BROKER_SOCKET:-}" ]; then
printf '%s\n' "$MOSAIC_LEASE_BROKER_SOCKET"
return 0
fi
local runtime_dir="${XDG_RUNTIME_DIR:-/run/user/$(id -u)}"
printf '%s\n' "${runtime_dir}/mosaic-lease/broker.sock"
}
if [ "$MODE" = "launch" ]; then
_broker_socket=$(broker_socket_path)
if [ ! -S "$_broker_socket" ]; then
echo "[fleet] FAIL_LAUNCH broker-absent: lease broker socket ${_broker_socket} missing; runtime launch denied (#1292)." >&2
echo "[fleet] remedy: systemctl --user enable --now mosaic-lease-broker.service (or reinstall via: mosaic fleet install)" >&2
exit 75
fi
fi
# Validate exact server ownership before querying, cleaning, or creating any
# managed session. An unmanaged or contaminated named socket is never repaired.
assert_owned_tmux_server
if [ "$MODE" = interaction ]; then
@@ -1,222 +0,0 @@
#!/usr/bin/env bash
# CI-fit regression suite for the #1292 lease-broker socket preflight in
# start-agent-session.sh.
#
# WHY THIS SUITE IS CI-FIT WHERE test-start-agent-session.sh IS NOT (#1017/#1270
# context): that older suite's precondition is "the host does not have the pi
# binary", which a CI image that ships pi violates — its guard correctly
# refuses to report a pass there, so it is excluded from the chain. THIS suite
# controls its own preconditions instead of inheriting them from the host: a
# fake tmux on PATH, a fake mosaic on PATH, a real unix socket created in a
# tmpdir, a hermetic env (env -i, fake HOME, GIT_CONFIG_GLOBAL severed). It
# never depends on what the host has installed, so a green here means the same
# thing on every host. Anyone adding cases: keep that property — no case may
# depend on host state.
#
# The failure this suite is written down to catch (#1292): a seat launched on a
# host with no lease broker dies ~4 seconds in at registration, with the
# diagnostic invisible because tmux destroys the dead pane. The preflight runs
# BEFORE any tmux effect and refuses with a NAMED code (exit 75, EX_TEMPFAIL)
# so the message survives. The agent@ unit is Type=oneshot with no Restart=,
# so a failed unit keeps its output instead of looping.
#
# Cases:
# 1. absent socket -> exit 75, message names broker-absent + socket path +
# remedy, and NO tmux session was ever created (the doomed-pane half).
# 2. present socket (real unix socket in tmpdir) -> proceeds PAST the
# preflight (the suite then stops at the next precondition, proving the
# preflight was not the refusal).
# 3. explicit MOSAIC_LEASE_BROKER_SOCKET wins over XDG_RUNTIME_DIR default.
# 4. --stop mode does NOT require the broker (teardown must not be fenced on
# a component whose absence is exactly what teardown may follow).
#
# Sabotage control, run by the developer (not in-suite): remove the preflight
# block from start-agent-session.sh, re-run — case 1 fails (a tmux session is
# created / exit is not 75), cases 2-4 still pass; restore byte-identically.
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/agent-session-broker-preflight}"
FAKE_HOME="$WORK_DIR/home"
BIN_DIR="$WORK_DIR/bin"
ENV_DIR="$WORK_DIR/env"
SOCK_DIR="$WORK_DIR/sockets"
LOG_FILE="$WORK_DIR/tmux-calls.log"
rm -rf "$WORK_DIR"
# The script asserts a managed directory tree under MOSAIC_HOME: mosaic/,
# mosaic/fleet/, mosaic/fleet/agents/ — private (0700/0750-style) modes, no
# symlinks — plus a per-agent env projection. Build the full tree the launcher
# expects so the suite reaches the BROKER preflight rather than dying at
# environment validation.
mkdir -p "$FAKE_HOME/.config/mosaic/fleet/agents" "$BIN_DIR" "$SOCK_DIR"
chmod 700 "$FAKE_HOME/.config/mosaic" "$FAKE_HOME/.config/mosaic/fleet/agents"
chmod 750 "$FAKE_HOME/.config/mosaic/fleet"
cat > "$FAKE_HOME/.config/mosaic/fleet/agents/preflight-test.env.generated" <<'ENVEOF'
MOSAIC_AGENT_NAME=preflight-test
MOSAIC_GIT_IDENTITY=preflight-test
MOSAIC_AGENT_CLASS=worker
MOSAIC_AGENT_RUNTIME=pi
MOSAIC_AGENT_MODEL=
MOSAIC_AGENT_REASONING=
MOSAIC_AGENT_TOOL_POLICY=code
MOSAIC_AGENT_WORKDIR=/tmp
MOSAIC_TMUX_SOCKET=mosaic-fleet
ENVEOF
chmod 600 "$FAKE_HOME/.config/mosaic/fleet/agents/preflight-test.env.generated"
# ─── Fake tmux: records every invocation; new-session marks the marker. ────
: > "$LOG_FILE"
cat > "$BIN_DIR/tmux" <<SH
#!/usr/bin/env bash
printf 'tmux %s\n' "\$*" >> "$LOG_FILE"
if [[ "\$*" == *new-session* ]]; then
echo "TMUX-NEW-SESSION-INVOKED" >> "$LOG_FILE"
fi
exit 0
SH
chmod +x "$BIN_DIR/tmux"
# ─── Fake mosaic/pi binaries so the script proceeds past its own lookups. ───
for bin in mosaic pi claude; do
printf '#!/usr/bin/env bash\nexit 0\n' > "$BIN_DIR/$bin"
chmod +x "$BIN_DIR/$bin"
done
# ─── Minimal launch environment the script expects. ────────────────────────
# (Enough for the preflight to be reached; later stages will still fail in
# case 2 — that is expected and asserted.)
run_session_script() {
local mode="$1"; shift
(
cd "$WORK_DIR"
env -i HOME="$FAKE_HOME" PATH="$BIN_DIR:/usr/bin:/bin" \
GIT_CONFIG_GLOBAL=/dev/null GIT_CONFIG_SYSTEM=/dev/null \
MOSAIC_HOME="$FAKE_HOME/.config/mosaic" \
AGENT_NAME=preflight-test \
"$@" \
bash "$SCRIPT_DIR/start-agent-session.sh" $mode preflight-test
)
}
fail=0
assert() {
local desc="$1" expected="$2" actual="$3"
if [[ "$expected" != "$actual" ]]; then
echo "FAIL: $desc — expected '$expected', got '$actual'" >&2
fail=1
fi
}
assert_contains() {
local desc="$1" haystack="$2" needle="$3"
[[ "$haystack" == *"$needle"* ]] || { echo "FAIL: $desc — missing '$needle' in: $haystack" >&2; fail=1; }
}
assert_not_contains() {
local desc="$1" haystack="$2" needle="$3"
if [[ "$haystack" == *"$needle"* ]]; then
echo "FAIL: $desc — must not contain '$needle'" >&2
fail=1
fi
return 0
}
# ─── 1. Absent socket → named refusal, NO tmux session. ────────────────────
: > "$LOG_FILE"
stderr_file="$WORK_DIR/stderr-1.tmp"
set +e
out=$(run_session_script "" MOSAIC_LEASE_BROKER_SOCKET="$SOCK_DIR/absent.sock" 2>"$stderr_file")
rc=$?
set -e
assert "absent socket exit code" "75" "$rc"
err=$(cat "$stderr_file")
assert_contains "absent socket names the failure" "$err" "FAIL_LAUNCH broker-absent"
assert_contains "absent socket names the socket path" "$err" "$SOCK_DIR/absent.sock"
assert_contains "absent socket names a remedy" "$err" "mosaic fleet install"
log1=$(cat "$LOG_FILE")
assert_not_contains "absent socket must not create a tmux session" "$log1" "TMUX-NEW-SESSION-INVOKED"
# ─── 2. Present socket → passes the preflight. ─────────────────────────────
# Expected: ownership/env checks AFTER the preflight may refuse (fixture is
# minimal by design); the assertion is only that the refusal is NOT
# broker-absent and the exit is NOT 75.
# Create a REAL unix socket: a detached python holder binds it and stays alive
# for the duration (bash cannot create sockets; a foreground python would
# close the socket on exit and -S on a closed-but-unlinked path fails). Written
# as a script file + setsid nohup so no job-control/heredoc interaction with
# set -e can silently kill the suite.
# AF_UNIX binds cap at 108 path bytes; the suite's workdir exceeds that, so
# the live socket lives at a SHORT path under /tmp (unique per run, cleaned
# with the suite). The preflight takes its socket path explicitly, so this
# stays fully controlled.
# A real unix socket at a SHORT absolute path (AF_UNIX limit is 108 bytes,
# so the repo-deep SOCK_DIR cannot host it). The name is composed, not
# `mktemp -u`: the CI image's mktemp dialect rejects that invocation
# (pipeline 2562: "mktemp: : Invalid argument"), and no pre-existing file is
# wanted anyway — the holder binds it fresh.
LIVE_SOCK="/tmp/mosaic-preflight-$RANDOM-$$.sock"
trap 'rm -f "$LIVE_SOCK"' EXIT
rm -f "$SOCK_DIR/live.sock" "$LIVE_SOCK"
cat > "$SOCK_DIR/holder.py" <<'PY'
import socket, sys, time
path = sys.argv[1]
s = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
s.bind(path)
s.listen(1)
time.sleep(120)
PY
python3 "$SOCK_DIR/holder.py" "$LIVE_SOCK" >/dev/null 2>"$SOCK_DIR/holder.err" &
HOLDER_PID=$!
# Wait for the socket object to exist (bind is near-instant, but do not race it).
for _ in $(seq 1 50); do
[ -S "$LIVE_SOCK" ] && break
sleep 0.1
done
if [ ! -S "$LIVE_SOCK" ]; then
echo "FAIL: could not create live socket fixture (holder pid $HOLDER_PID)" >&2
ps -p "$HOLDER_PID" -o pid,stat,cmd --no-headers >&2 || echo "(holder exited)" >&2
cat "$SOCK_DIR/holder.err" >&2 || true
exit 1
fi
: > "$LOG_FILE"
set +e
out=$(run_session_script "" MOSAIC_LEASE_BROKER_SOCKET="$LIVE_SOCK" 2>"$WORK_DIR/stderr-2.tmp")
rc=$?
set -e
# The preflight PASSED if the failure (whatever later stage refused) is NOT
# the broker refusal, and tmux was reached or a later precondition named
# something else.
err2=$(cat "$WORK_DIR/stderr-2.tmp")
assert_not_contains "live socket must not refuse broker-absent" "$err2" "broker-absent"
if [[ "$rc" == "75" ]]; then
echo "FAIL: live socket — preflight still refused (exit 75) with a live socket" >&2
fail=1
fi
# ─── 3. Explicit socket env wins over XDG default. ─────────────────────────
set +e
out=$(run_session_script "" XDG_RUNTIME_DIR="$SOCK_DIR/no-runtime-here" MOSAIC_LEASE_BROKER_SOCKET="$SOCK_DIR/absent2.sock" 2>"$WORK_DIR/stderr-3.tmp")
rc=$?
set -e
assert "explicit env wins (exit 75)" "75" "$rc"
assert_contains "explicit env path named" "$(cat "$WORK_DIR/stderr-3.tmp")" "$SOCK_DIR/absent2.sock"
# ─── 4. --stop is not fenced on the broker. ────────────────────────────────
: > "$LOG_FILE"
set +e
out=$(run_session_script "--stop" MOSAIC_LEASE_BROKER_SOCKET="$SOCK_DIR/absent3.sock" 2>"$WORK_DIR/stderr-4.tmp")
rc=$?
set -e
err4=$(cat "$WORK_DIR/stderr-4.tmp")
assert_not_contains "--stop must not refuse broker-absent" "$err4" "broker-absent"
if [[ "$rc" == "75" ]]; then
echo "FAIL: --stop — exit 75 means teardown was fenced on the broker" >&2
fail=1
fi
kill "$HOLDER_PID" 2>/dev/null || true
if [[ "$fail" -eq 0 ]]; then
echo "start-agent-session lease-broker preflight regression passed"
fi
exit "$fail"
+1 -59
View File
@@ -30,9 +30,7 @@ The Gitea API token is **never passed on a curl command line.** An `Authorizatio
### `--login` override
Both `pr-review.sh` and `issue-comment.sh` accept an optional `--login <name>` flag that overrides the automatically detected Gitea login for that single invocation. The override selects **which credential the REST write, the `/user` identity lookup, and the read-back all use** — its token is resolved from the tea config for that login name (`get_gitea_token_for_login`).
**With no `--login`, there is no tea lookup at all.** The acting credential is the calling identity's own, resolved by `get_gitea_token` (see "Per-agent Gitea identity" below), and there is deliberately no fallback from it. These wrappers previously _guessed_ a login from the repo host and looked that guess up in the tea config; on a shared-account host the guess resolved to the shared login, so an unqualified call authored its write as that account rather than as the caller. Since `get_gitea_token_for_login` matches by login **name** and performs no authentication check, a dead shared credential still resolved at rc=0 and the identity-aware resolver was never reached. A caller passing no `--login` is asking to act as itself, so `--login` is now the only route to the tea store (#1351). The resolved login is **host- and port-bound**: the login's configured URL host **and effective port** (the scheme's default port — 80 for `http`, 443 for `https` — applies when a port is omitted, symmetrically on both sides) must match the repo remote's, so a login name shared across hosts (or an override configured for a different Gitea, including one on a different port of the same host) can never send one host's credential to another — a host or port mismatch fails closed rather than leaking a cross-host token. Resolving the acting identity and the read-back from the _same_ login that performs the write is essential: a write performed under an overridden login must be verified against that login's identity, not the host default's. Callers who need a different login than the host default should pass `--login <reviewer-login>`.
Both `pr-review.sh` and `issue-comment.sh` accept an optional `--login <name>` flag that overrides the automatically detected Gitea login for that single invocation. The override selects **which credential the REST write, the `/user` identity lookup, and the read-back all use** — its token is resolved from the tea config for that login name (`get_gitea_token_for_login`), falling back to the repo host's credential when no login is named. The resolved login is **host- and port-bound**: the login's configured URL host **and effective port** (the scheme's default port — 80 for `http`, 443 for `https` — applies when a port is omitted, symmetrically on both sides) must match the repo remote's, so a login name shared across hosts (or an override configured for a different Gitea, including one on a different port of the same host) can never send one host's credential to another — a host or port mismatch fails closed rather than leaking a cross-host token. Resolving the acting identity and the read-back from the _same_ login that performs the write is essential: a write performed under an overridden login must be verified against that login's identity, not the host default's. Callers who need a different login than the host default should pass `--login <reviewer-login>`.
As a durable successor to this mechanism, consider giving each reviewer/approver slot its own dedicated Gitea login credential, so that author≠reviewer holds at the credential level rather than relying on wrapper-level `--login` bookkeeping. This is a recommendation for future hardening, not something implemented by this flag.
@@ -102,36 +100,6 @@ of their own — `MOSAIC_GIT_IDENTITY=<id>` with a provisioned slot. There is de
environment variable that restores the fallback; one would reintroduce exactly the
substitution this removes.
### The tea path: login resolution (#1356)
The wrappers that go through `tea` (`issue-list.sh`, `pr-list.sh`, `pr-view.sh`,
`lane-brief.sh`, and the tea half of `issue-close.sh`) cannot use a token directly: tea
0.14 only acts as a **login** already stored in `~/.config/tea/config.yml`. Those wrappers
therefore resolve a login name, not a token, and the resolution follows the same identity
as above:
1. Resolve the identity (`MOSAIC_GIT_IDENTITY`, then `git config mosaic.gitIdentity`).
2. Derive the Gitea instance from the repo host (`git.mosaicstack.dev``mosaicstack`,
`git.uscllc.com``usc`), or from the owner when `--repo owner/name` is given.
3. The canonical login is `<instance>-<identity>`. If tea has it, that login acts.
4. If the identity is set but that login is missing, the wrapper **fails closed**: nonzero
exit, empty stdout, and a stderr line naming the login it wanted and the source of the
identity. When `tea` itself is not installed the message says so instead, since "no such
login" would send the reader to create a login they cannot create.
5. With **no identity set**, the old host-default behaviour is unchanged (first login
configured for that host, else the API fallback).
Step 4 replaced a fallback that picked any login configured for the host, which meant a
seat with no login of its own silently acted as whichever seat had configured one. That
satisfied the author≠reviewer gate on paper while one actor held both names.
**Provisioning the logins.** `tools/fleet/seat-logins.sh` projects each seat's token from
its secrets store into tea's config under the canonical name. tea's config is a derived
cache of the secrets store: regenerate it with the script, never hand-edit it. Run it with
`--seat <seat>` for one seat (all seats when omitted), dry-run by default, `--apply` to write. A hand-made
alias holding a seat's token blocks its canonical name (tea refuses one token under two
names); `--adopt` renames it.
### Enabling it for a clone
The framework installer syncs `git-credential-mosaic` to
@@ -157,32 +125,6 @@ otherwise careful never to touch. Because identity is already resolved per-workt
(`mosaic.gitIdentity`), the correct granularity for registering the helper is per-clone
too, so a documented manual step is the right shape here, not a global auto-write.
### Running these tests
`MOSAIC_GIT_IDENTITY` is inherited into each test's sandbox `HOME`, and **the tests disagree
about which value they need**, so no single ambient value passes all 29. Measured on `next` at
`a480ee83`, two full passes differing only in that variable:
| tests | identity exported | identity unset |
| ----------------------------------------------------------------------------------------------------------------------------------------------- | ----------------- | -------------- |
| `gitea-login-resolution`, `issue-comment-readback`, `issue-create-interactive-auth`, `pr-edit`, `pr-merge-gitea-empty-uid`, `pr-metadata-gitea` | **fail** | pass |
| `issue-close-fail-closed` | pass | **fail** |
| remaining 22 | pass | pass |
The six fail because inside a sandbox `HOME` the identity has no `fleet/agents/<id>/`
directory, so it is classified as a **service identity**, its store is unpopulated, and the
resolver correctly refuses with `Refusing to borrow another slot's token`. That is the
documented fail-closed behaviour above, reached from a state the test never intended.
`issue-close-fail-closed` is the mirror image: it asserts that no comment POST is attempted, so
it needs an identity resolving to an empty slot, and with the variable unset the shared account
answers and the POST goes through.
These read as wrapper regressions rather than as environment. Two seats independently
misdiagnosed them as a patch defect while reviewing #1352. Until each test controls its own
value (#1353), `env -u MOSAIC_GIT_IDENTITY` is the closest thing to a clean run at 28/29, with
`issue-close-fail-closed` the expected failure — and **"the suite passes" is not a statement
anyone can make here without naming the ambient value that produced it.**
### PowerShell parity
`detect-platform.ps1`'s Gitea wrappers authenticate through `tea` CLI logins
@@ -1,6 +1,6 @@
#!/bin/bash
# ci-queue-wait.sh - Wait until project CI queue is clear (no running/queued pipeline on branch head)
# Usage: ci-queue-wait.sh [-B branch] [-t timeout_sec] [-i interval_sec] [--purpose push|merge] [--require-status] [--no-ci-expected]
# Usage: ci-queue-wait.sh [-B branch] [-t timeout_sec] [-i interval_sec] [--purpose push|merge] [--require-status]
set -euo pipefail
@@ -14,11 +14,10 @@ TIMEOUT_SEC=900
INTERVAL_SEC=15
PURPOSE="merge"
REQUIRE_STATUS=0
NO_CI_EXPECTED=0
usage() {
cat <<EOF
Usage: $(basename "$0") [-B branch] [-R owner/repo] [--sha full-40] [-t timeout_sec] [-i interval_sec] [--purpose push|merge] [--require-status] [--no-ci-expected]
Usage: $(basename "$0") [-B branch] [-R owner/repo] [--sha full-40] [-t timeout_sec] [-i interval_sec] [--purpose push|merge] [--require-status]
Options:
-B, --branch BRANCH Branch head to inspect (default: current branch)
@@ -28,7 +27,6 @@ Options:
-i, --interval SECONDS Poll interval in seconds (default: 15)
--purpose VALUE Log context: push|merge (default: merge)
--require-status Fail if no CI status contexts are present
--no-ci-expected Assert this repository has no CI configured: a merge guard on a zero-context head becomes queue-clear (requires the acting token to hold repository admin); refused with exit 78 when MOSAIC_GIT_IDENTITY is unset or empty
-h, --help Show this help
Examples:
@@ -177,50 +175,6 @@ PY
return 0
}
# Durable audit record for an explicit no-CI assertion event (granted or
# refused). Same JSONL sink and field shape as record_cannot_assert so one
# reader covers all three outcomes; the outcome value distinguishes them.
# rc 70 on an unwritable sink: a merge pass that cannot be audited must not
# be reachable, mirroring record_cannot_assert's refusal of a degraded pass.
record_assertion_event() {
local outcome="$1" reason="$2" asserted_by="$3"
local audit_log="${MOSAIC_CI_QUEUE_AUDIT_LOG:-${XDG_STATE_HOME:-${HOME:-}/.local/state}/mosaic/audit/ci-queue-wait.jsonl}"
if [[ -z "$audit_log" ]] || ! mkdir -p "$(dirname "$audit_log")"; then
echo "Error: could not write ${outcome} audit record (audit directory unavailable at ${audit_log})." >&2
return 70
fi
if ! python3 - "$audit_log" "$outcome" "$reason" "$asserted_by" "${PLATFORM:-unknown}" "$PURPOSE" "${BRANCH:-unknown}" "${OWNER:-unknown}/${REPO:-unknown}" <<'PY'
import datetime
import json
import os
import sys
path, outcome, reason, asserted_by, platform, purpose, branch, repo = sys.argv[1:]
record = {
"timestamp": datetime.datetime.now(datetime.timezone.utc).isoformat(),
"outcome": outcome,
"reason": reason,
"platform": platform,
"purpose": purpose,
"branch": branch,
"repo": repo,
"asserted_by": asserted_by,
}
fd = os.open(path, os.O_WRONLY | os.O_CREAT | os.O_APPEND, 0o600)
try:
os.write(fd, (json.dumps(record, separators=(",", ":")) + "\n").encode())
finally:
os.close(fd)
PY
then
echo "Error: could not write ${outcome} audit record at ${audit_log}; refusing to proceed unaudited." >&2
return 70
fi
return 0
}
github_get_branch_head_sha() {
local owner="$1"
local repo="$2"
@@ -228,24 +182,6 @@ github_get_branch_head_sha() {
gh api "repos/${owner}/${repo}/branches/${branch}" --jq '.commit.sha'
}
# Repository-admin state for the acting credential, GitHub flavor. The
# repository object's permissions.admin is the field; read through the same
# gh CLI the guard already authenticates with. rc 0 = admin, 1 = not admin
# (or field absent), 2 = indeterminate (transport/API failure).
github_repo_admin_state() {
local owner="$1"
local repo="$2"
local perm
if ! perm=$(gh api "repos/${owner}/${repo}" --jq '.permissions.admin' 2>/dev/null); then
return 2
fi
case "$perm" in
true) return 0 ;;
false|null|"") return 1 ;;
*) return 2 ;;
esac
}
github_get_commit_status_json() {
local owner="$1"
local repo="$2"
@@ -370,41 +306,6 @@ gitea_get_commit_status_json() {
curl -fsSL -H "User-Agent: curl/8" -H "Authorization: token ${token}" "$url"
}
# Repository-admin state for the acting credential, Gitea flavor. The guard's
# existing fetches (branch head, combined status) carry no permissions object
# (measured: neither response includes one), so the elevation check reads the
# repository object's permissions.admin, the one documented carrier of that
# field. rc 0 = admin, 1 = not admin (or field absent), 2 = indeterminate
# (non-200 or unparseable).
gitea_repo_admin_state() {
local host="$1"
local repo="$2"
local token="$3"
local url="https://${host}/api/v1/repos/${repo}"
local resp code body
resp=$(curl -sS -H "User-Agent: curl/8" -H "Authorization: token ${token}" -w $'\n%{http_code}' "$url") || return 2
code="${resp##*$'\n'}"
body="${resp%$'\n'*}"
if [[ "$code" != "200" ]]; then
return 2
fi
printf '%s' "$body" | python3 -c '
import json
import sys
try:
payload = json.load(sys.stdin)
except Exception:
raise SystemExit(2)
if not isinstance(payload, dict):
raise SystemExit(2)
permissions = payload.get("permissions")
if not isinstance(permissions, dict) or permissions.get("admin") is not True:
raise SystemExit(1)
raise SystemExit(0)
'
}
while [[ $# -gt 0 ]]; do
case "$1" in
-B|--branch)
@@ -435,10 +336,6 @@ while [[ $# -gt 0 ]]; do
REQUIRE_STATUS=1
shift
;;
--no-ci-expected)
NO_CI_EXPECTED=1
shift
;;
-h|--help)
usage
exit 0
@@ -468,10 +365,6 @@ if [[ "$PURPOSE" != "push" && "$PURPOSE" != "merge" ]]; then
echo "Error: --purpose must be push or merge." >&2
exit 1
fi
if [[ "$NO_CI_EXPECTED" -eq 1 && "$REQUIRE_STATUS" -eq 1 ]]; then
echo "Error: --no-ci-expected and --require-status contradict each other: one asserts the repository has no CI, the other demands status contexts. Pass at most one." >&2
exit 1
fi
OWNER="unknown"
REPO="unknown"
@@ -591,48 +484,6 @@ while true; do
echo "[ci-queue-wait] queue-clear state=no-status purpose=push branch=${BRANCH}; no queued or running CI."
exit 0
fi
if [[ "$NO_CI_EXPECTED" -eq 1 ]]; then
# Explicit, elevated, audit-visible assertion that this
# repository has no CI to wait on. The zero-context case is
# the ONLY state the flag reclassifies: a pending or failed
# context still holds or fails exactly as without it, and a
# non-admin token is refused rather than trusted.
# The assertion must name an asserting identity: "unknown"
# attributes nothing, so a caller with MOSAIC_GIT_IDENTITY
# unset or empty is refused (exit 78) BEFORE the permission
# lookup -- an unattributable caller never triggers that
# network call.
if [[ -z "${MOSAIC_GIT_IDENTITY:-}" ]]; then
record_assertion_event "ASSERTION_UNATTRIBUTABLE" "actor-unattributable" "unknown" \
|| echo "Warning: could not write the ASSERTION_UNATTRIBUTABLE audit record; the refusal itself stands." >&2
echo "Error: ASSERTION_UNATTRIBUTABLE state=no-status purpose=merge asserted-by=unknown reason=no-ci-expected branch=${BRANCH}; --no-ci-expected requires MOSAIC_GIT_IDENTITY to name the asserting identity and it is unset or empty (exit 78)." >&2
exit 78
fi
ASSERTED_BY="${MOSAIC_GIT_IDENTITY}"
ADMIN_STATE=2
if [[ "$PLATFORM" == "github" ]]; then
if github_repo_admin_state "$OWNER" "$REPO"; then ADMIN_STATE=0; else ADMIN_STATE=$?; fi
else
if gitea_repo_admin_state "$HOST" "$OWNER/$REPO" "$TOKEN"; then ADMIN_STATE=0; else ADMIN_STATE=$?; fi
fi
case "$ADMIN_STATE" in
0)
record_assertion_event "NO_CI_ASSERTED" "no-ci-expected" "$ASSERTED_BY" || exit $?
echo "[ci-queue-wait] queue-clear state=no-status purpose=merge asserted-by=${ASSERTED_BY} reason=no-ci-expected branch=${BRANCH}"
exit 0
;;
1)
record_assertion_event "ASSERTION_REFUSED" "actor-not-repo-admin" "$ASSERTED_BY" \
|| echo "Warning: could not write the ASSERTION_REFUSED audit record; the refusal itself stands." >&2
echo "Error: ASSERTION_REFUSED state=no-status purpose=merge asserted-by=${ASSERTED_BY} reason=no-ci-expected branch=${BRANCH}; --no-ci-expected requires repository admin and the acting token is not an admin of ${OWNER}/${REPO} (exit 77)." >&2
exit 77
;;
*)
record_cannot_assert "repo-permissions-unavailable"
exit $?
;;
esac
fi
echo "Error: ASSERTED_NOT_READY state=no-status purpose=${PURPOSE} branch=${BRANCH}." >&2
exit 3
;;
@@ -180,66 +180,6 @@ raise SystemExit(1)
PY
}
# Map a host to the instance prefix used in canonical tea login names
# ("<instance>-<identity>"). This deliberately mirrors the _idpfx case in
# get_gitea_token(): the two credential paths must agree on what a host is called,
# or an agent authenticates as itself on one path and as somebody else on the other.
gitea_instance_for_host() {
case "${1:-}" in
git.uscllc.com) echo usc ;;
git.mosaicstack.dev) echo mosaicstack ;;
*) return 1 ;;
esac
}
# Resolve the acting git identity, same precedence as get_gitea_token() step 0.
# Prints "<identity>\t<source>" so the caller can name the source in an error.
resolve_git_identity() {
local ident src
ident="${MOSAIC_GIT_IDENTITY:-}"
src="MOSAIC_GIT_IDENTITY"
if [[ -z "$ident" ]]; then
ident="$(git config --get mosaic.gitIdentity 2>/dev/null || true)"
src="git config mosaic.gitIdentity"
fi
[[ -n "$ident" ]] || return 1
printf '%s\t%s\n' "$ident" "$src"
}
# Map a repo owner to an instance. Used only by the --repo override path, which
# has an owner and no host. Previously lived inline in lane-brief.sh; one copy so
# the two override callers cannot drift apart.
gitea_instance_for_owner() {
local owner="${1:-}"
owner="${owner%%/*}"
case "$owner" in
usc|USC) echo usc ;;
mosaicstack|mosaic) echo mosaicstack ;;
*) return 1 ;;
esac
}
# Does a login of this name exist at all? The --repo override path cannot check
# host agreement, because it has no host.
tea_login_exists() {
local login_name="$1"
local logins_json
command -v tea >/dev/null 2>&1 || return 1
logins_json=$(tea login list --output json 2>/dev/null) || return 1
TEA_LOGINS_JSON="$logins_json" python3 - "$login_name" <<'PY_INNER'
import json, os, sys
want = sys.argv[1]
try:
logins = json.loads(os.environ.get("TEA_LOGINS_JSON", "[]"))
except Exception:
raise SystemExit(1)
for login in logins if isinstance(logins, list) else []:
if str(login.get("name") or login.get("Name") or "") == want:
raise SystemExit(0)
raise SystemExit(1)
PY_INNER
}
tea_login_matches_host() {
local login_name="$1" host="$2"
local logins_json
@@ -336,40 +276,6 @@ get_gitea_login_for_host() {
fi
fi
# IDENTITY LADDER (#1356). Below this point the old code took the FIRST login
# matching the host, which is not an identity — with 43 logins on a fleet host,
# ~22 match one server, so a seat with no login of its own silently acted as
# whichever happened to be first in ~/.config/tea/config.yml. Gate 16 depends on
# author != reviewer, and borrowing satisfies it mechanically while violating it
# in fact. The token path already refuses to borrow; this is the same refusal.
#
# Enforced ONLY when an identity is resolvable, exactly like get_gitea_token():
# no identity means a human at a terminal, and neither path enforces there.
local ident ident_src inst canon
if IFS=$'\t' read -r ident ident_src < <(resolve_git_identity); then
if inst=$(gitea_instance_for_host "$host"); then
canon="${inst}-${ident}"
if tea_login_matches_host "$canon" "$host"; then
echo "$canon"
return 0
fi
# Say which of the two it is. "No such login" when tea is simply not
# installed is a diagnosis of a cause that was never checked, and it
# sends the reader off to create a login they cannot create.
if ! command -v tea >/dev/null 2>&1; then
echo "Error: git identity '$ident' requested (via $ident_src) for host '$host', but tea is not installed," >&2
echo " so no login can be resolved. Refusing to guess an identity." >&2
return 1
fi
echo "Error: git identity '$ident' requested (via $ident_src) for host '$host', but no tea login named '$canon' exists." >&2
echo " Refusing to borrow another login. Acting as a different identity would satisfy gate 16 mechanically while violating it." >&2
echo " Create it with: ~/.config/mosaic/tools/fleet/seat-logins.sh --apply --seat $ident" >&2
return 1
fi
# Identity known but the host is not a Mosaic instance. Fall through: the
# canonical name is undefined for it, so there is nothing to enforce.
fi
login=$(find_tea_login_for_host "$host" || true)
if [[ -n "$login" ]]; then
echo "$login"
@@ -445,49 +351,14 @@ raise SystemExit(1)
PY
}
# Resolve a login for an explicit --repo override, which supplies an owner and no
# host. Takes "owner" or "owner/repo".
#
# The old body fell through to get_default_tea_login(), which returns the
# default-marked login or, failing that, the first login of ANY host — arbitrary
# identity, chosen by config file order. That is the #1356 fail-open in its worst
# form, because unlike the host path it does not even constrain the server.
get_gitea_login_for_repo_override() {
local owner="${1:-}"
local login ident ident_src inst canon
local login
if [[ -n "${GITEA_LOGIN:-}" ]]; then
echo "$GITEA_LOGIN"
return 0
fi
if IFS=$'\t' read -r ident ident_src < <(resolve_git_identity); then
if inst=$(gitea_instance_for_owner "$owner"); then
canon="${inst}-${ident}"
if tea_login_exists "$canon"; then
echo "$canon"
return 0
fi
# Same split as the host path above (#1357 S1): a missing tea binary
# is not a missing login, and the "create it with" advice cannot be
# followed without tea.
if ! command -v tea >/dev/null 2>&1; then
echo "Error: git identity '$ident' (via $ident_src) requested for owner '${owner%%/*}', but tea is not installed," >&2
echo " so no login can be resolved. Refusing to guess an identity." >&2
return 1
fi
echo "Error: git identity '$ident' (via $ident_src) has no tea login '$canon' for owner '${owner%%/*}'." >&2
echo " Create it with: ~/.config/mosaic/tools/fleet/seat-logins.sh --apply --seat $ident" >&2
return 1
fi
echo "Error: git identity '$ident' (via $ident_src) is set, but owner '${owner%%/*}' maps to no known instance," >&2
echo " so the login name cannot be derived. Refusing to fall back to an arbitrary login." >&2
echo " Set GITEA_LOGIN to name the login explicitly." >&2
return 1
fi
# No identity: a human at a terminal. Unchanged, and the same place the token
# path stops enforcing.
login=$(get_default_tea_login || true)
if [[ -n "$login" ]]; then
echo "$login"
@@ -102,17 +102,9 @@ gitea_resolve_api_for_login() {
return 1
}
else
# NO --login: the acting credential is this identity's own token and there
# is deliberately no tea-config fallback. get_gitea_token_for_login matches
# by login NAME and performs no authentication check, and with no --login
# that name was a HOST GUESS resolving to a shared account. A live shared
# token would therefore have authored every seat's comment as that
# account, making Gate-16 author-is-not-reviewer unenforceable fleet-wide;
# a dead one is only what made the defect visible. get_gitea_token fails
# loud on a fleet host when no identity resolves, and that refusal is the
# correct outcome, not a case to fall back from.
GITEA_API_TOKEN=$(get_gitea_token "$host") || {
echo "Error: no Gitea credential resolved for the acting identity on host '$host' (comment write/read-back). Set MOSAIC_GIT_IDENTITY=<agent-id>, or pass --login <name> to use a named tea credential." >&2
GITEA_API_TOKEN=$(get_gitea_token_for_login "$effective_login" "$host") \
|| GITEA_API_TOKEN=$(get_gitea_token "$host") || {
echo "Error: Gitea token not found for login '$effective_login' (comment write/read-back)" >&2
return 1
}
fi
@@ -343,12 +335,15 @@ if [[ "$PLATFORM" == "github" ]]; then
gh issue comment "$ISSUE_NUMBER" --body "$COMMENT"
echo "Added comment to GitHub issue #$ISSUE_NUMBER"
elif [[ "$PLATFORM" == "gitea" ]]; then
# A --login override selects a NAMED tea credential and is the only way to
# reach the tea store. With no --login there is deliberately no guess: the
# comment is attributed to this identity's own credential, resolved by
# gitea_resolve_api_for_login. The guess this replaced named a SHARED
# account, selecting an identity the caller never asked to act as.
# Resolve the login this comment should be attributed to: the --login
# override when given, otherwise the detected default for this repo's host.
# A --login override always wins. Otherwise name this repo host's login only
# as a best effort: the login name merely selects a per-login token, and
# gitea_resolve_api_for_login falls back to the host credential
# (get_gitea_token) when no tea login is named, so the default credential
# still resolves even when the host tea has no matching login entry.
EFFECTIVE_LOGIN="$LOGIN_OVERRIDE"
[[ -n "$EFFECTIVE_LOGIN" ]] || EFFECTIVE_LOGIN=$(get_gitea_login 2>/dev/null || true)
# Bind the REST endpoint + token to the effective login, then derive the
# acting identity from that SAME credential (GET /user). The write below and
@@ -99,8 +99,8 @@ case "$PLATFORM" in
;;
gitea)
if [[ -n "$REPO_OVERRIDE" ]]; then
GITEA_LOGIN_NAME=$(get_gitea_login_for_repo_override "$REPO_OVERRIDE") || {
echo "Error: could not resolve a Gitea login for the --repo override (the lines above say why). Set GITEA_LOGIN to name one explicitly." >&2
GITEA_LOGIN_NAME=$(get_gitea_login_for_repo_override) || {
echo "Error: Could not resolve Gitea login for --repo override. Set GITEA_LOGIN or configure a default tea login." >&2
exit 1
}
else
@@ -1,5 +1,5 @@
#!/bin/bash
# issue-view.sh - View issue details, including comments, on GitHub or Gitea
# issue-view.sh - View issue details on GitHub or Gitea
# Usage: issue-view.sh -i <issue_number>
set -e
@@ -28,47 +28,11 @@ gitea_issue_view_api() {
}
url="https://${host}/api/v1/repos/${repo}/issues/${ISSUE_NUMBER}"
local -a curl_args=(-fsS -H "User-Agent: curl/8" -H "Authorization: token ${token}")
if ! command -v python3 >/dev/null 2>&1; then
# No renderer: raw JSON is all this path can give. Comments are a
# second resource, so fetch them too rather than only the count.
curl "${curl_args[@]}" "$url"
curl "${curl_args[@]}" "${url}/comments"
return
if command -v python3 >/dev/null 2>&1; then
curl -fsS -H "User-Agent: curl/8" -H "Authorization: token ${token}" "$url" | python3 -m json.tool
else
curl -fsS -H "User-Agent: curl/8" -H "Authorization: token ${token}" "$url"
fi
# Render issue + comments as text (#1357 F2). The old fallback dumped the
# issue JSON, which carries only a comment COUNT, so every comment body was
# invisible on this path and the wrapper could never show what
# `tea issues --comments` shows.
{
curl "${curl_args[@]}" "$url"
echo
echo "__MOSAIC_COMMENTS__"
curl "${curl_args[@]}" "${url}/comments"
} | python3 -c '
import json, sys
raw = sys.stdin.read()
issue_raw, _, comments_raw = raw.partition("__MOSAIC_COMMENTS__")
issue = json.loads(issue_raw)
comments = json.loads(comments_raw) if comments_raw.strip() else []
print("#%s %s" % (issue["number"], issue["title"]))
print("State: %s Author: %s Created: %s" % (issue["state"], issue["user"]["login"], issue["created_at"]))
labels = ", ".join(l["name"] for l in issue.get("labels") or [])
if labels:
print("Labels: " + labels)
if issue.get("milestone"):
print("Milestone: " + issue["milestone"]["title"])
print("URL: " + issue["html_url"])
print()
print(issue.get("body") or "(no body)")
if comments:
print()
print("--- Comments (%d) ---" % len(comments))
for c in comments:
print()
print("[%s at %s]" % (c["user"]["login"], c["created_at"]))
print(c.get("body") or "")
'
}
while [[ $# -gt 0 ]]; do
@@ -82,8 +46,6 @@ while [[ $# -gt 0 ]]; do
echo ""
echo "Options:"
echo " -i, --issue Issue number (required)"
echo ""
echo "Comments are always included (tea --comments / Gitea API /comments)."
echo " -h, --help Show this help"
exit 0
;;
@@ -105,30 +67,11 @@ if [[ "$PLATFORM" == "github" ]]; then
gh issue view "$ISSUE_NUMBER"
elif [[ "$PLATFORM" == "gitea" ]]; then
if command -v tea >/dev/null 2>&1; then
# --comments is what makes tea print the comment bodies (#1357 F3).
# Without it tea prompts for them interactively, which in a
# non-interactive wrapper means they are silently never shown.
tea_err=$(mktemp)
if tea issue "$ISSUE_NUMBER" $(get_gitea_repo_args) --comments 2>"$tea_err"; then
rm -f "$tea_err"
if tea issue "$ISSUE_NUMBER" $(get_gitea_repo_args); then
exit 0
fi
# Name the cause tea actually reported, not a guessed one (#1357 F1/F4).
# tea reads the cwd's git config before honouring --repo; a repo with
# extensions.worktreeconfig=true makes it exit 1 with a
# repositoryformatversion error. That is a git-config condition, not a
# credential one. The old path printed the REVOKED OR STALE TOKEN note
# here unconditionally, which sent readers to rotate a token that was fine.
if grep -q 'repositoryformatversion' "$tea_err"; then
echo "Warning: tea cannot read this repo's git config (extensions.worktreeconfig); not a credential problem. Using Gitea API fallback." >&2
elif grep -q 'user does not exist' "$tea_err"; then
echo "Warning: tea issue view failed, trying Gitea API fallback..." >&2
{ declare -F explain_tea_user_does_not_exist >/dev/null && explain_tea_user_does_not_exist; } || true
else
echo "Warning: tea issue view failed, trying Gitea API fallback..." >&2
fi
sed 's/^/ tea: /' "$tea_err" >&2
rm -f "$tea_err"
echo "Warning: tea issue view failed, trying Gitea API fallback..." >&2
{ declare -F explain_tea_user_does_not_exist >/dev/null && explain_tea_user_does_not_exist; } || true
fi
gitea_issue_view_api
else
@@ -49,27 +49,11 @@ if [[ -z "$LOGIN" ]]; then
if [[ -n "${GITEA_LOGIN:-}" ]]; then
LOGIN="$GITEA_LOGIN"
else
# #1356: the owner-derived map below picks a SHARED login (bare `usc` /
# `mosaicstack`). On a seat that is borrowing another identity, which is
# exactly what gate 16 forbids. So the identity ladder goes first and the
# map is only the no-identity fallback (a human at a terminal), which is
# where the token path stops enforcing too.
if LOGIN="$(get_gitea_login_for_repo_override "$REPO")"; then
:
elif resolve_git_identity >/dev/null 2>&1; then
# A git identity IS set and the ladder still could not resolve a login.
# The named reason is already on stderr. Falling through to the map here
# would hand this seat a SHARED login (bare `usc` / `mosaicstack`) — the
# identity-borrowing #1356 exists to stop. Fail closed instead.
exit 2
else
# No identity: a human at a terminal. Owner-derived map, unchanged. This
# is the same point at which the token path stops enforcing.
case "${REPO%%/*}" in
usc|USC) LOGIN=usc ;;
mosaicstack|mosaic) LOGIN=mosaicstack ;;
esac
fi
case "${REPO%%/*}" in
usc|USC) LOGIN=usc ;;
mosaicstack|mosaic) LOGIN=mosaicstack ;;
*) LOGIN="$(get_gitea_login_for_repo_override 2>/dev/null || true)" ;;
esac
fi
fi
[[ -n "$LOGIN" ]] || { echo "FATAL: could not resolve a Gitea login for $REPO (pass -L or set GITEA_LOGIN)" >&2; exit 2; }
@@ -19,41 +19,6 @@ ISSUE=""
# get_remote_host, get_gitea_token, get_repo_info, and get_gitea_repo_args are provided by detect-platform.sh
gitea_default_branch() {
# Forge default branch for the current repo (T51-P2 WP5a / spec E4): the
# API fallback must not guess a base. Empty output or any lookup failure
# returns nonzero so the caller fails loud instead of mistargeting a PR.
local host repo token url body branch
host=$(get_remote_host) || return 1
repo=$(get_repo_info) || return 1
token=$(get_gitea_token "$host") || return 1
url="https://${host}/api/v1/repos/${repo}"
# Fetch and parse as separate steps (T51P2WP5AR B2): a piped
# `curl | python` reports only python's status, so an HTTP failure that
# still emits parseable JSON would masquerade as success. curl's own
# exit status is authoritative here.
if ! body=$(curl -fsS \
-H "User-Agent: curl/8" \
-H "Authorization: token ${token}" \
"$url" 2>/dev/null); then
return 1
fi
# A valid base is a NONBLANK JSON STRING (T51P2WP5AR B3): null, numbers,
# and whitespace-only values are failed resolution, never a POSTed base.
branch=$(printf '%s' "$body" | python3 -c '
import json, sys
try:
value = json.load(sys.stdin).get("default_branch")
except Exception:
sys.exit(1)
if not isinstance(value, str) or not value.strip():
sys.exit(1)
print(value.strip())
' 2>/dev/null) || return 1
[[ -n "$branch" ]] || return 1
printf '%s' "$branch"
}
gitea_pr_create_api() {
local host repo token url payload
host=$(get_remote_host) || {
@@ -73,28 +38,14 @@ gitea_pr_create_api() {
echo "Warning: API fallback applies title/body/head/base only; labels/milestone/draft require authenticated tea setup." >&2
fi
# Base resolution (spec E4): an explicit -B always wins; with none, the
# forge default branch is resolved from the provider API -- never the
# historical "main" literal, which mistargeted every fallback PR on
# repos whose trunk is not main (e.g. mosaicstack/stack -> next).
local api_base=""
if [[ -n "$BASE_BRANCH" ]]; then
api_base="$BASE_BRANCH"
else
api_base=$(gitea_default_branch) || {
echo "Error: could not resolve the forge default branch for the API-fallback base; pass -B <branch> explicitly" >&2
return 1
}
fi
payload=$(TITLE="$TITLE" BODY="$BODY" HEAD_BRANCH="$HEAD_BRANCH" API_BASE="$api_base" python3 - <<'PY'
payload=$(TITLE="$TITLE" BODY="$BODY" HEAD_BRANCH="$HEAD_BRANCH" BASE_BRANCH="$BASE_BRANCH" python3 - <<'PY'
import json
import os
payload = {
"title": os.environ["TITLE"],
"head": os.environ["HEAD_BRANCH"],
"base": os.environ["API_BASE"],
"base": os.environ["BASE_BRANCH"] or "main",
}
body = os.environ.get("BODY", "")
if body:
@@ -121,7 +72,7 @@ Create a pull request on the current repository (Gitea or GitHub).
Options:
-t, --title TITLE PR title (required, or use --issue)
-b, --body BODY PR description/body
-B, --base BRANCH Base branch to merge into (default: the forge repository's default branch)
-B, --base BRANCH Base branch to merge into (default: main/master)
-H, --head BRANCH Head branch with changes (default: current branch)
-l, --labels LABELS Comma-separated labels
-m, --milestone NAME Milestone name
@@ -94,8 +94,8 @@ case "$PLATFORM" in
;;
gitea)
if [[ -n "$REPO_OVERRIDE" ]]; then
GITEA_LOGIN_NAME=$(get_gitea_login_for_repo_override "$REPO_OVERRIDE") || {
echo "Error: could not resolve a Gitea login for the --repo override (the lines above say why). Set GITEA_LOGIN to name one explicitly." >&2
GITEA_LOGIN_NAME=$(get_gitea_login_for_repo_override) || {
echo "Error: Could not resolve Gitea login for --repo override. Set GITEA_LOGIN or configure a default tea login." >&2
exit 1
}
else
@@ -1,6 +1,6 @@
#!/bin/bash
# pr-merge.sh - Merge pull requests on Gitea or GitHub
# Usage: pr-merge.sh -n PR_NUMBER [-m squash] [-d] [--expect-head SHA] [--no-ci-expected] [--co-author-trailers --escalate-to PRINCIPAL]
# Usage: pr-merge.sh -n PR_NUMBER [-m squash] [-d] [--expect-head SHA] [--co-author-trailers --escalate-to PRINCIPAL]
set -euo pipefail
@@ -16,7 +16,6 @@ DRY_RUN=false
EXPECT_HEAD=""
CO_AUTHOR_TRAILERS=false
ESCALATE_TO=""
NO_CI_EXPECTED=false
usage() {
cat <<EOF
@@ -30,7 +29,6 @@ Options:
-d, --delete-branch Delete the head branch after merge
--dry-run Run metadata/login preflight without merging
--expect-head SHA Refuse unless the PR head matches this full commit SHA
--no-ci-expected Assert the target repository has no CI: forward --no-ci-expected to the queue guard (requires repository admin)
--co-author-trailers Build verified trailers from linked PR commit authors
--escalate-to NAME Named principal for an unresolved-author BLOCK
-h, --help Show this help message
@@ -72,10 +70,6 @@ while [[ $# -gt 0 ]]; do
EXPECT_HEAD="$2"
shift 2
;;
--no-ci-expected)
NO_CI_EXPECTED=true
shift
;;
--co-author-trailers)
CO_AUTHOR_TRAILERS=true
shift
@@ -160,18 +154,13 @@ if [[ "$DRY_RUN" != true ]]; then
if [[ -z "$BASE_REPO" ]]; then
BASE_REPO="$(get_repo_owner)/$(get_repo_name)"
fi
guard_args=(
--purpose merge
-B "$HEAD_BRANCH"
-R "$BASE_REPO"
--sha "$HEAD_SHA"
-t "${MOSAIC_CI_QUEUE_TIMEOUT_SEC:-900}"
"$SCRIPT_DIR/ci-queue-wait.sh" \
--purpose merge \
-B "$HEAD_BRANCH" \
-R "$BASE_REPO" \
--sha "$HEAD_SHA" \
-t "${MOSAIC_CI_QUEUE_TIMEOUT_SEC:-900}" \
-i "${MOSAIC_CI_QUEUE_POLL_SEC:-15}"
)
if [[ "$NO_CI_EXPECTED" == true ]]; then
guard_args+=(--no-ci-expected)
fi
"$SCRIPT_DIR/ci-queue-wait.sh" "${guard_args[@]}"
fi
PLATFORM=$(detect_platform)
@@ -12,7 +12,7 @@
# concurrent record cannot masquerade as this write and a no-op fails closed.
#
# --login override: the default login is resolved from the local tea login list
# for this repo's host from the acting identity's own credential. Pass --login <name> to
# for this repo's host (get_gitea_login_for_host). Pass --login <name> to
# override it for this invocation only. The REST write, the /user identity read,
# and every read-back are ALL performed with the token of the EFFECTIVE login,
# so the write and its verification bind to the same identity.
@@ -372,17 +372,9 @@ gitea_resolve_api_for_login() {
return 1
}
else
# NO --login: the acting credential is this identity's own token and there
# is deliberately no tea-config fallback. get_gitea_token_for_login matches
# by login NAME and performs no authentication check, and with no --login
# that name was a HOST GUESS resolving to a shared account. A live shared
# token would therefore have authored every seat's review as that
# account, making Gate-16 author-is-not-reviewer unenforceable fleet-wide;
# a dead one is only what made the defect visible. get_gitea_token fails
# loud on a fleet host when no identity resolves, and that refusal is the
# correct outcome, not a case to fall back from.
GITEA_API_TOKEN=$(get_gitea_token "$host") || {
echo "Error: no Gitea credential resolved for the acting identity on host '$host' (review write/read-back). Set MOSAIC_GIT_IDENTITY=<agent-id>, or pass --login <name> to use a named tea credential." >&2
GITEA_API_TOKEN=$(get_gitea_token_for_login "$effective_login" "$host") \
|| GITEA_API_TOKEN=$(get_gitea_token "$host") || {
echo "Error: Gitea token not found for login '$effective_login' (review write/read-back)" >&2
return 1
}
fi
@@ -706,7 +698,7 @@ if [[ "$PLATFORM" == "github" ]]; then
elif [[ "$PLATFORM" == "gitea" ]]; then
case $ACTION in
approve)
# Best-effort host for credential resolution only (gitea_resolve_api_for_login
# Best-effort host for the tea-login GUESS only (gitea_resolve_api_for_login
# below re-derives the real host from HOST_OVERRIDE/remote independently and
# is authoritative). Prefer an explicit -H/--host; otherwise best-effort
# git-remote inference, tolerating its ABSENCE (a bare `get_remote_host` here
@@ -714,13 +706,15 @@ elif [[ "$PLATFORM" == "gitea" ]]; then
# SILENTLY — exit 1, zero output — even though -r/-H are exactly the flags
# that support running with no usable origin at all).
host="${HOST_OVERRIDE:-$(get_remote_host 2>/dev/null || true)}"
# A --login override selects a NAMED tea credential and is the only
# way to reach the tea store. With no --login there is deliberately no
# guess: gitea_resolve_api_for_login resolves this identity's own token.
# The guess this replaced named a SHARED account, selecting an identity
# the caller never asked to act as. The single resolved token is then
# used for the write, the /user identity, and the read-back.
# A --login override always wins. Otherwise name this host's login
# only as a best effort: the login name merely selects a per-login
# token, and gitea_resolve_api_for_login falls back to the host
# credential (get_gitea_token) when no tea login is named — so a host
# tea's login list need not enumerate exotic (e.g. ported) hosts for
# the default credential to resolve. The single resolved token is
# then used for the write, the /user identity, and the read-back.
EFFECTIVE_LOGIN="$LOGIN_OVERRIDE"
[[ -n "$EFFECTIVE_LOGIN" ]] || EFFECTIVE_LOGIN=$(get_gitea_login_for_host "$host" 2>/dev/null || true)
# Bind the REST endpoint + token to the effective login, then derive
# the acting identity from that SAME credential so the review submit
# and its read-back verify against the identity that performed them.
@@ -741,7 +735,7 @@ elif [[ "$PLATFORM" == "gitea" ]]; then
echo "Error: Comment required for request-changes"
exit 1
fi
# Best-effort host for credential resolution only (gitea_resolve_api_for_login
# Best-effort host for the tea-login GUESS only (gitea_resolve_api_for_login
# below re-derives the real host from HOST_OVERRIDE/remote independently and
# is authoritative). Prefer an explicit -H/--host; otherwise best-effort
# git-remote inference, tolerating its ABSENCE (a bare `get_remote_host` here
@@ -749,13 +743,15 @@ elif [[ "$PLATFORM" == "gitea" ]]; then
# SILENTLY — exit 1, zero output — even though -r/-H are exactly the flags
# that support running with no usable origin at all).
host="${HOST_OVERRIDE:-$(get_remote_host 2>/dev/null || true)}"
# A --login override selects a NAMED tea credential and is the only
# way to reach the tea store. With no --login there is deliberately no
# guess: gitea_resolve_api_for_login resolves this identity's own token.
# The guess this replaced named a SHARED account, selecting an identity
# the caller never asked to act as. The single resolved token is then
# used for the write, the /user identity, and the read-back.
# A --login override always wins. Otherwise name this host's login
# only as a best effort: the login name merely selects a per-login
# token, and gitea_resolve_api_for_login falls back to the host
# credential (get_gitea_token) when no tea login is named — so a host
# tea's login list need not enumerate exotic (e.g. ported) hosts for
# the default credential to resolve. The single resolved token is
# then used for the write, the /user identity, and the read-back.
EFFECTIVE_LOGIN="$LOGIN_OVERRIDE"
[[ -n "$EFFECTIVE_LOGIN" ]] || EFFECTIVE_LOGIN=$(get_gitea_login_for_host "$host" 2>/dev/null || true)
gitea_resolve_api_for_login "$EFFECTIVE_LOGIN" "${LOGIN_OVERRIDE:+explicit}" || exit 1
ACTING_LOGIN=$(gitea_authenticated_login) || exit 1
head_sha=$(gitea_pr_head_sha "$PR_NUMBER") || exit 1
@@ -770,7 +766,7 @@ elif [[ "$PLATFORM" == "gitea" ]]; then
echo "Error: Comment required"
exit 1
fi
# Best-effort host for credential resolution only (gitea_resolve_api_for_login
# Best-effort host for the tea-login GUESS only (gitea_resolve_api_for_login
# below re-derives the real host from HOST_OVERRIDE/remote independently and
# is authoritative). Prefer an explicit -H/--host; otherwise best-effort
# git-remote inference, tolerating its ABSENCE (a bare `get_remote_host` here
@@ -778,13 +774,15 @@ elif [[ "$PLATFORM" == "gitea" ]]; then
# SILENTLY — exit 1, zero output — even though -r/-H are exactly the flags
# that support running with no usable origin at all).
host="${HOST_OVERRIDE:-$(get_remote_host 2>/dev/null || true)}"
# A --login override selects a NAMED tea credential and is the only
# way to reach the tea store. With no --login there is deliberately no
# guess: gitea_resolve_api_for_login resolves this identity's own token.
# The guess this replaced named a SHARED account, selecting an identity
# the caller never asked to act as. The single resolved token is then
# used for the write, the /user identity, and the read-back.
# A --login override always wins. Otherwise name this host's login
# only as a best effort: the login name merely selects a per-login
# token, and gitea_resolve_api_for_login falls back to the host
# credential (get_gitea_token) when no tea login is named — so a host
# tea's login list need not enumerate exotic (e.g. ported) hosts for
# the default credential to resolve. The single resolved token is
# then used for the write, the /user identity, and the read-back.
EFFECTIVE_LOGIN="$LOGIN_OVERRIDE"
[[ -n "$EFFECTIVE_LOGIN" ]] || EFFECTIVE_LOGIN=$(get_gitea_login_for_host "$host" 2>/dev/null || true)
gitea_resolve_api_for_login "$EFFECTIVE_LOGIN" "${LOGIN_OVERRIDE:+explicit}" || exit 1
ACTING_LOGIN=$(gitea_authenticated_login) || exit 1
comment_id=$(gitea_create_comment_verified "$PR_NUMBER" "$COMMENT" "$ACTING_LOGIN") || {
@@ -59,8 +59,8 @@ if [[ "$PLATFORM" == "github" ]]; then
gh pr view "$PR_NUMBER" --repo "$REPO_INFO"
elif [[ "$PLATFORM" == "gitea" ]]; then
if [[ -n "$REPO_OVERRIDE" ]]; then
GITEA_LOGIN_NAME=$(get_gitea_login_for_repo_override "$REPO_OVERRIDE") || {
echo "Error: could not resolve a Gitea login for the --repo override (the lines above say why). Set GITEA_LOGIN to name one explicitly." >&2
GITEA_LOGIN_NAME=$(get_gitea_login_for_repo_override) || {
echo "Error: Could not resolve Gitea login for --repo override. Set GITEA_LOGIN or configure a default tea login." >&2
exit 1
}
else
@@ -1,323 +0,0 @@
#!/usr/bin/env bash
# Regression harness for ci-queue-wait.sh's --no-ci-expected assertion:
# the sanctioned merge path for a repository with no CI configured at all.
#
# Zero status contexts ("no-status") stays fail-closed for --purpose merge
# by default, because at merge time no-status can also mean "CI has not
# reported yet". --no-ci-expected reclassifies ONLY that zero-context case
# as queue-clear, and only for a caller whose acting token holds repository
# admin. This harness pins:
# (a) merge + no-status + flag + admin -> exit 0, audit line + JSONL.
# (b) merge + no-status, no flag -> exit 3, existing text (unchanged).
# (c) merge + no-status + flag + non-admin -> exit 77 ASSERTION_REFUSED
# (distinct text, exit code NOT 3) + JSONL refusal record.
# (c2) flag + admin payload without the admin field -> fail closed as (c).
# (d) flag + --require-status -> usage error, before any network.
# (e) flag + a real pending context -> still holds (timeout 124),
# and the admin endpoint is never consulted.
# (f) push + no-status, with and without the flag -> push queue-clear
# unchanged; no admin consultation on push.
# (g) flag + admin lookup unreachable -> CANNOT_ASSERT hold (75),
# not a silent pass and not a refusal.
# (h) flag + admin stub + NO MOSAIC_GIT_IDENTITY -> refusal BEFORE
# queue-clear and BEFORE the admin lookup: exit 78, no queue-clear
# line, an ASSERTION_UNATTRIBUTABLE JSONL record, no repos/ call.
set -u
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/ci-queue-wait-no-ci-expected}"
REPO_DIR="$WORK_DIR/repo"
STUB_DIR="$WORK_DIR/stubs"
URL_LOG="$WORK_DIR/urls.log"
rm -rf "$WORK_DIR"
mkdir -p "$REPO_DIR" "$STUB_DIR"
git -C "$REPO_DIR" init -q
git -C "$REPO_DIR" remote add origin https://git.example.test/acme/widgets.git
# Same stub conventions as test-ci-queue-wait-no-status.sh; adds the
# repository-object endpoint (admin state) selected by MOSAIC_STUB_ADMIN_MODE.
cat > "$STUB_DIR/curl" <<'SH'
#!/usr/bin/env bash
set -euo pipefail
has_w=0
url=""
for arg in "$@"; do
case "$arg" in
-w) has_w=1 ;;
http://*|https://*) url="$arg" ;;
esac
done
printf '%s\n' "$url" >> "${MOSAIC_STUB_URL_LOG:?}"
case "$url" in
*/branches/*)
body='{"commit":{"id":"deadbeefcafef00d0123456789abcdef01234567"}}'
if [[ "$has_w" == 1 ]]; then
printf '%s\n200' "$body"
else
printf '%s' "$body"
fi
exit 0
;;
*/status)
mode="${MOSAIC_STUB_STATUS_MODE:?MOSAIC_STUB_STATUS_MODE not set}"
case "$mode" in
no-status) body='{"state":"","statuses":[]}' ;;
real-pending) body='{"state":"pending","statuses":[{"context":"ci/woodpecker","status":"running","target_url":""}]}' ;;
*) echo "curl stub: unknown status mode=$mode" >&2; exit 2 ;;
esac
printf '%s' "$body"
exit 0
;;
*/repos/*)
mode="${MOSAIC_STUB_ADMIN_MODE:?MOSAIC_STUB_ADMIN_MODE not set}"
case "$mode" in
admin) body='{"permissions":{"admin":true,"push":true,"pull":true}}' ;;
non-admin) body='{"permissions":{"admin":false,"push":true,"pull":true}}' ;;
no-admin-field) body='{"permissions":{}}' ;;
unreachable) exit 7 ;;
*) echo "curl stub: unknown admin mode=$mode" >&2; exit 2 ;;
esac
if [[ "$has_w" == 1 ]]; then
printf '%s\n200' "$body"
else
printf '%s' "$body"
fi
exit 0
;;
*)
echo "curl stub: unrecognized URL: $url" >&2
exit 2
;;
esac
SH
chmod +x "$STUB_DIR/curl"
failures=0
run_guard() {
local name="$1"; shift
(
cd "$REPO_DIR" || exit
export PATH="$STUB_DIR:$PATH"
export MOSAIC_CREDENTIALS_FILE="$WORK_DIR/no-credentials.json"
export MOSAIC_CI_QUEUE_AUDIT_LOG="$WORK_DIR/audit-$name.jsonl"
export MOSAIC_STUB_URL_LOG="$URL_LOG"
export GITEA_TOKEN="stub-token"
export GITEA_URL="https://git.example.test"
export MOSAIC_GIT_IDENTITY="test-identity"
"$SCRIPT_DIR/ci-queue-wait.sh" -B main -t 3 -i 1 "$@"
)
}
# The suite exports test-identity globally, so the unattributable-caller
# case must strip it from the child environment at invocation with env -u,
# not rely on the export order.
run_guard_no_identity() {
local name="$1"; shift
(
cd "$REPO_DIR" || exit
export PATH="$STUB_DIR:$PATH"
export MOSAIC_CREDENTIALS_FILE="$WORK_DIR/no-credentials.json"
export MOSAIC_CI_QUEUE_AUDIT_LOG="$WORK_DIR/audit-$name.jsonl"
export MOSAIC_STUB_URL_LOG="$URL_LOG"
export GITEA_TOKEN="stub-token"
export GITEA_URL="https://git.example.test"
export MOSAIC_GIT_IDENTITY="test-identity"
env -u MOSAIC_GIT_IDENTITY \
"$SCRIPT_DIR/ci-queue-wait.sh" -B main -t 3 -i 1 "$@"
)
}
expect_rc() {
local name="$1" want="$2" got="$3"
if [[ "$want" == "not3" ]]; then
if [[ "$got" -eq 0 || "$got" -eq 3 ]]; then
echo "FAIL $name: expected a refusal rc (nonzero, not 3), got $got" >&2
failures=$((failures + 1))
return 1
fi
elif [[ "$got" -ne "$want" ]]; then
echo "FAIL $name: expected rc=$want, got rc=$got" >&2
failures=$((failures + 1))
return 1
fi
return 0
}
expect_text() {
local name="$1" want="$2" output="$3" polarity="${4:-present}"
if [[ "$polarity" == "present" && "$output" != *"$want"* ]]; then
echo "FAIL $name: output missing '$want'" >&2
printf '%s\n' "$output" >&2
failures=$((failures + 1))
elif [[ "$polarity" == "absent" && "$output" == *"$want"* ]]; then
echo "FAIL $name: output unexpectedly contains '$want'" >&2
printf '%s\n' "$output" >&2
failures=$((failures + 1))
fi
}
repo_root_fetched() {
grep -q 'repos/acme/widgets$' "$URL_LOG"
}
# (a) merge + no-status + flag + admin -> exit 0, assertion line, JSONL record.
: > "$URL_LOG"
set +e
out_a=$(MOSAIC_STUB_STATUS_MODE=no-status MOSAIC_STUB_ADMIN_MODE=admin run_guard a --purpose merge --no-ci-expected 2>&1)
rc_a=$?
set -u
if expect_rc a 0 "$rc_a"; then
expect_text a "queue-clear state=no-status purpose=merge asserted-by=test-identity reason=no-ci-expected branch=main" "$out_a"
expect_text a "ASSERTED_NOT_READY" "$out_a" absent
if ! grep -q '"outcome":"NO_CI_ASSERTED"' "$WORK_DIR/audit-a.jsonl" 2>/dev/null; then
echo "FAIL a: expected a NO_CI_ASSERTED JSONL audit record" >&2
failures=$((failures + 1))
elif ! grep -q '"asserted_by":"test-identity"' "$WORK_DIR/audit-a.jsonl"; then
echo "FAIL a: audit record does not name the asserting identity" >&2
failures=$((failures + 1))
fi
fi
# (b) merge + no-status, no flag -> exit 3, existing error text unchanged.
: > "$URL_LOG"
set +e
out_b=$(MOSAIC_STUB_STATUS_MODE=no-status run_guard b --purpose merge 2>&1)
rc_b=$?
set -u
if expect_rc b 3 "$rc_b"; then
expect_text b "Error: ASSERTED_NOT_READY state=no-status purpose=merge branch=main." "$out_b"
expect_text b "asserted-by" "$out_b" absent
fi
if repo_root_fetched; then
echo "FAIL b: admin endpoint consulted without the flag" >&2
failures=$((failures + 1))
fi
# (c) merge + no-status + flag + non-admin -> distinct refusal, rc NOT 3.
: > "$URL_LOG"
set +e
out_c=$(MOSAIC_STUB_STATUS_MODE=no-status MOSAIC_STUB_ADMIN_MODE=non-admin run_guard c --purpose merge --no-ci-expected 2>&1)
rc_c=$?
set -u
if expect_rc c not3 "$rc_c"; then
if [[ "$rc_c" -ne 77 ]]; then
echo "FAIL c: expected the documented refusal rc=77, got $rc_c" >&2
failures=$((failures + 1))
fi
expect_text c "ASSERTION_REFUSED state=no-status purpose=merge asserted-by=test-identity reason=no-ci-expected branch=main" "$out_c"
expect_text c "ASSERTED_NOT_READY" "$out_c" absent
if ! grep -q '"outcome":"ASSERTION_REFUSED"' "$WORK_DIR/audit-c.jsonl" 2>/dev/null; then
echo "FAIL c: expected an ASSERTION_REFUSED JSONL audit record" >&2
failures=$((failures + 1))
fi
fi
# (c2) admin payload with no admin field -> fail closed as non-admin.
: > "$URL_LOG"
set +e
out_c2=$(MOSAIC_STUB_STATUS_MODE=no-status MOSAIC_STUB_ADMIN_MODE=no-admin-field run_guard c2 --purpose merge --no-ci-expected 2>&1)
rc_c2=$?
set -u
if expect_rc c2 77 "$rc_c2"; then
expect_text c2 "ASSERTION_REFUSED" "$out_c2"
fi
# (d) flag + --require-status -> usage error before any network I/O.
: > "$URL_LOG"
set +e
out_d=$(MOSAIC_STUB_STATUS_MODE=no-status MOSAIC_STUB_ADMIN_MODE=admin run_guard d --purpose merge --no-ci-expected --require-status 2>&1)
rc_d=$?
set -u
if expect_rc d 1 "$rc_d"; then
expect_text d "--no-ci-expected and --require-status contradict" "$out_d"
fi
if [[ -s "$URL_LOG" ]]; then
echo "FAIL d: usage error must precede every network call" >&2
failures=$((failures + 1))
fi
# (e) flag + a real pending context -> still holds; admin endpoint never asked.
: > "$URL_LOG"
set +e
out_e=$(MOSAIC_STUB_STATUS_MODE=real-pending MOSAIC_STUB_ADMIN_MODE=admin run_guard e --purpose merge --no-ci-expected 2>&1)
rc_e=$?
set -u
if expect_rc e 124 "$rc_e"; then
expect_text e "ASSERTED_NOT_READY" "$out_e"
expect_text e "ci/woodpecker=running" "$out_e"
fi
if repo_root_fetched; then
echo "FAIL e: a pending context must not trigger the admin assertion" >&2
failures=$((failures + 1))
fi
# (f) push + no-status stays queue-clear, with and without the flag.
: > "$URL_LOG"
set +e
out_f=$(MOSAIC_STUB_STATUS_MODE=no-status MOSAIC_STUB_ADMIN_MODE=non-admin run_guard f --purpose push 2>&1)
rc_f=$?
set -u
if expect_rc f 0 "$rc_f"; then
expect_text f "queue-clear state=no-status purpose=push branch=main; no queued or running CI." "$out_f"
fi
: > "$URL_LOG"
set +e
out_f2=$(MOSAIC_STUB_STATUS_MODE=no-status MOSAIC_STUB_ADMIN_MODE=non-admin run_guard f2 --purpose push --no-ci-expected 2>&1)
rc_f2=$?
set -u
if expect_rc f2 0 "$rc_f2"; then
expect_text f2 "queue-clear state=no-status purpose=push branch=main; no queued or running CI." "$out_f2"
expect_text f2 "asserted-by" "$out_f2" absent
fi
if repo_root_fetched; then
echo "FAIL f: push must not consult the admin endpoint" >&2
failures=$((failures + 1))
fi
# (g) flag + admin lookup unreachable -> CANNOT_ASSERT hold (75), not a pass.
: > "$URL_LOG"
set +e
out_g=$(MOSAIC_STUB_STATUS_MODE=no-status MOSAIC_STUB_ADMIN_MODE=unreachable run_guard g --purpose merge --no-ci-expected 2>&1)
rc_g=$?
set -u
if expect_rc g 75 "$rc_g"; then
expect_text g "CANNOT_ASSERT reason=repo-permissions-unavailable" "$out_g"
fi
if ! grep -q '"outcome":"CANNOT_ASSERT"' "$WORK_DIR/audit-g.jsonl" 2>/dev/null; then
echo "FAIL g: expected a CANNOT_ASSERT JSONL audit record" >&2
failures=$((failures + 1))
fi
# (h) flag + admin stub + no asserting identity -> refusal before queue-clear
# and before the admin lookup: rc 78, no queue-clear line, an
# ASSERTION_UNATTRIBUTABLE JSONL record, and zero repos/ network calls.
: > "$URL_LOG"
set +e
out_h=$(MOSAIC_STUB_STATUS_MODE=no-status MOSAIC_STUB_ADMIN_MODE=admin run_guard_no_identity h --purpose merge --no-ci-expected 2>&1)
rc_h=$?
set -u
if expect_rc h 78 "$rc_h"; then
expect_text h "ASSERTION_UNATTRIBUTABLE state=no-status purpose=merge asserted-by=unknown reason=no-ci-expected branch=main" "$out_h"
expect_text h "queue-clear" "$out_h" absent
if ! grep -q '"outcome":"ASSERTION_UNATTRIBUTABLE"' "$WORK_DIR/audit-h.jsonl" 2>/dev/null; then
echo "FAIL h: expected an ASSERTION_UNATTRIBUTABLE JSONL audit record" >&2
failures=$((failures + 1))
fi
fi
if repo_root_fetched; then
echo "FAIL h: an unattributable caller must not trigger the permission lookup" >&2
failures=$((failures + 1))
fi
if [[ "$failures" -ne 0 ]]; then
echo "ci-queue-wait no-ci-expected regression failed ($failures assertions)" >&2
exit 1
fi
echo "ci-queue-wait no-ci-expected regression passed (all outcome classes)"
@@ -100,15 +100,6 @@ case "$url" in
*/commits/*/status)
printf '{"state":"success","statuses":[{"context":"ci/mock","status":"success"}]}'
;;
# Repo roots: the pr-create API fallback resolves its base from the forge
# default_branch (T51-P2 WP5a). Exact-suffix matches so the /pulls POST
# endpoint (no trailing path) still falls through to the catch-all.
*/api/v1/repos/USC/uconnect)
printf '{"default_branch":"main"}'
;;
*/api/v1/repos/mosaicstack/stack)
printf '{"default_branch":"next"}'
;;
*)
printf '{}'
;;
@@ -119,20 +110,8 @@ chmod +x "$BIN_DIR/tea" "$BIN_DIR/curl"
run_in_repo() {
(
# HERMETICITY, second half (#1356). The empty repo-local `mosaic.gitIdentity`
# above pins the git-config route into identity resolution. It does NOT pin
# the environment route, and MOSAIC_GIT_IDENTITY is checked FIRST — so on any
# provisioned seat, where the launcher exports it, this suite failed before
# any change: rc=1 as-is, rc=0 under `env -u MOSAIC_GIT_IDENTITY`, one
# variable. A suite that cannot run on a seat cannot guard this code for the
# agents that actually run it.
#
# Unset rather than set empty: an empty MOSAIC_GIT_IDENTITY and an absent one
# take different branches in resolve_git_identity(), and the case under test
# is "no identity at all".
cd "$REPO_DIR"
env -u MOSAIC_GIT_IDENTITY \
PATH="${_SANDBOX_BIN:-$BIN_DIR}:$PATH" \
PATH="$BIN_DIR:$PATH" \
HOME="$HOME_DIR" \
MOSAIC_CREDENTIALS_FILE="$CREDENTIALS_FILE" \
MOSAIC_TEST_LOG="$LOG_FILE" \
@@ -328,11 +307,14 @@ SH
chmod +x "$BIN_DIR2/tea"
run_in_repo2() {
# Same sandbox as run_in_repo, different mock tea (BIN_DIR2 defines a
# mosaicstack login). This MUST delegate rather than re-implement: it was a
# copy once, and the copy silently missed the MOSAIC_GIT_IDENTITY unset, so
# the suite kept failing on a seat after run_in_repo was already fixed.
_SANDBOX_BIN="$BIN_DIR2" run_in_repo "$@"
(
cd "$REPO_DIR"
PATH="$BIN_DIR2:$PATH" \
HOME="$HOME_DIR" \
MOSAIC_CREDENTIALS_FILE="$CREDENTIALS_FILE" \
MOSAIC_TEST_LOG="$LOG_FILE" \
"$@"
)
}
git -C "$REPO_DIR" remote set-url origin https://git.mosaicstack.dev/mosaicstack/stack.git
@@ -358,151 +340,6 @@ if [[ "$override_wins" != "mosaicstack" ]]; then
fi
git -C "$REPO_DIR" remote set-url origin https://git.uscllc.com/USC/uconnect.git
# ---------------------------------------------------------------------------
# #1356: the git-identity ladder. A seat declares who it is (MOSAIC_GIT_IDENTITY
# or `git config mosaic.gitIdentity`); resolution must use THAT seat's login and
# must REFUSE to borrow another one when it is absent. Silently borrowing
# satisfies gate 16 mechanically (a review exists) while violating it (the
# reviewer and the author are the same actor under two names).
#
# BIN_DIR3 mocks a tea that holds a canonical per-seat login, which is what a
# projected seat looks like. BIN_DIR2 (mosaicstack only) is reused as the
# "seat has no login" case — no third mock needed for the negative branch.
# ---------------------------------------------------------------------------
BIN_DIR3="$WORK_DIR/bin3"
mkdir -p "$BIN_DIR3"
cp "$BIN_DIR/curl" "$BIN_DIR3/curl"
cat > "$BIN_DIR3/tea" <<'SH'
#!/usr/bin/env bash
set -euo pipefail
if [[ "$*" == "login list --output json" ]]; then
cat <<'JSON'
[
{"name":"mosaicstack","url":"https://git.mosaicstack.dev","user":"ci-bot"},
{"name":"mosaicstack-testseat","url":"https://git.mosaicstack.dev","user":"testseat"},
{"name":"usc","url":"https://git.uscllc.com","user":"ci-bot"}
]
JSON
exit 0
fi
printf 'tea %s\n' "$*" >> "$MOSAIC_TEST_LOG"
exit 0
SH
chmod +x "$BIN_DIR3/tea"
run_in_repo3() { _SANDBOX_BIN="$BIN_DIR3" run_in_repo "$@"; }
git -C "$REPO_DIR" remote set-url origin https://git.mosaicstack.dev/mosaicstack/stack.git
# Branch 1 (host path): identity set, canonical login PRESENT -> that login wins
# over the shared `mosaicstack` one, which is what host-matching alone would pick.
ladder_hit=$(run_in_repo3 env MOSAIC_GIT_IDENTITY=testseat bash -c '
source "'"$SCRIPT_DIR"'/detect-platform.sh"
get_gitea_login_for_host git.mosaicstack.dev
')
if [[ "$ladder_hit" != "mosaicstack-testseat" ]]; then
echo "Expected identity ladder to select 'mosaicstack-testseat'; got '$ladder_hit'" >&2
exit 1
fi
# CONTROL for branch 1: the same mock, no identity, must still resolve by host.
# Without this, branch 1 passing proves nothing about the ladder specifically --
# it would also pass if the code just picked the last matching login.
ladder_none=$(run_in_repo3 bash -c '
source "'"$SCRIPT_DIR"'/detect-platform.sh"
get_gitea_login_for_host git.mosaicstack.dev
')
if [[ "$ladder_none" != "mosaicstack" ]]; then
echo "Expected no-identity host resolution to stay 'mosaicstack'; got '$ladder_none'" >&2
exit 1
fi
# Branch 2 (host path): identity set, canonical login ABSENT -> fail closed with a
# named error. Two assertions, and they are not the same one twice: rc!=0 proves
# it refused, and the ABSENCE of any login on stdout proves it did not borrow the
# `mosaicstack` login that is sitting right there matching the host.
ladder_err=$(run_in_repo2 env MOSAIC_GIT_IDENTITY=testseat bash -c '
source "'"$SCRIPT_DIR"'/detect-platform.sh"
get_gitea_login_for_host git.mosaicstack.dev
' 2>&1 1>/dev/null || true)
ladder_out=$(run_in_repo2 env MOSAIC_GIT_IDENTITY=testseat bash -c '
source "'"$SCRIPT_DIR"'/detect-platform.sh"
get_gitea_login_for_host git.mosaicstack.dev
' 2>/dev/null || true)
if [[ -n "$ladder_out" ]]; then
echo "Identity ladder BORROWED login '$ladder_out' instead of failing closed" >&2
exit 1
fi
if ! grep -q "mosaicstack-testseat" <<<"$ladder_err"; then
echo "Expected the error to name the login it wanted; got: $ladder_err" >&2
exit 1
fi
# Branch 3: `git config mosaic.gitIdentity` is the second rung and must work when
# the environment variable is absent -- a seat may be configured either way.
git -C "$REPO_DIR" config mosaic.gitIdentity testseat
ladder_gitcfg=$(run_in_repo3 bash -c '
source "'"$SCRIPT_DIR"'/detect-platform.sh"
get_gitea_login_for_host git.mosaicstack.dev
')
git -C "$REPO_DIR" config --unset mosaic.gitIdentity || true
if [[ "$ladder_gitcfg" != "mosaicstack-testseat" ]]; then
echo "Expected git-config identity rung to select 'mosaicstack-testseat'; got '$ladder_gitcfg'" >&2
exit 1
fi
# Branch 4 (--repo override path): same rule, owner-derived instead of host-derived.
override_ladder=$(run_in_repo3 env MOSAIC_GIT_IDENTITY=testseat bash -c '
source "'"$SCRIPT_DIR"'/detect-platform.sh"
get_gitea_login_for_repo_override mosaicstack/stack
')
if [[ "$override_ladder" != "mosaicstack-testseat" ]]; then
echo "Expected --repo override ladder to select 'mosaicstack-testseat'; got '$override_ladder'" >&2
exit 1
fi
# Branch 5: explicit GITEA_LOGIN outranks the ladder. An operator naming a login
# by hand is a deliberate act, not an accident to be second-guessed.
override_explicit=$(run_in_repo3 env MOSAIC_GIT_IDENTITY=testseat GITEA_LOGIN=mosaicstack bash -c '
source "'"$SCRIPT_DIR"'/detect-platform.sh"
get_gitea_login_for_repo_override mosaicstack/stack
')
if [[ "$override_explicit" != "mosaicstack" ]]; then
echo "Expected explicit GITEA_LOGIN to outrank the identity ladder; got '$override_explicit'" >&2
exit 1
fi
# Branch 6 (#1357 S1): with tea ABSENT from PATH, the override path must say tea is
# missing, not "no tea login named X exists" (a cause that was never checked) and
# not the seat-logins.sh advice, which cannot be followed without tea.
NOTEA_BIN="$WORK_DIR/notea-bin"; mkdir -p "$NOTEA_BIN"
for t in bash git python3 sed grep cat mktemp dirname basename readlink env sort head tr cut; do
_p="$(command -v "$t" 2>/dev/null || true)"; [[ -n "$_p" ]] && ln -sf "$_p" "$NOTEA_BIN/$t"
done
override_notea_rc=0
override_notea_err=$(cd "$REPO_DIR" && env -u GITEA_LOGIN \
PATH="$NOTEA_BIN" HOME="$HOME_DIR" MOSAIC_GIT_IDENTITY=testseat \
bash -c '
command -v tea >/dev/null 2>&1 && { echo "SETUP: tea still on PATH"; exit 99; }
source "'"$SCRIPT_DIR"'/detect-platform.sh"
get_gitea_login_for_repo_override mosaicstack/stack
' 2>&1 >/dev/null) || override_notea_rc=$?
if [[ "$override_notea_rc" != 1 ]]; then
echo "Expected --repo override path to fail (rc=1) with tea absent; got rc=$override_notea_rc: $override_notea_err" >&2
exit 1
fi
if ! grep -q 'tea is not installed' <<<"$override_notea_err"; then
echo "Expected --repo override path to name tea as absent; got: $override_notea_err" >&2
exit 1
fi
if grep -q 'has no tea login\|seat-logins.sh' <<<"$override_notea_err"; then
echo "Override path diagnosed a missing LOGIN while tea itself is absent: $override_notea_err" >&2
exit 1
fi
git -C "$REPO_DIR" remote set-url origin https://git.uscllc.com/USC/uconnect.git
# ---------------------------------------------------------------------------
# #865 Blocker 1 & 2: get_gitea_token_for_login must resolve the SAME token as
# PyYAML would (or fail closed identically) even when PyYAML is ABSENT, and must
@@ -76,22 +76,7 @@ exit 0
EOF
chmod +x "$MOCK_BIN/tea"
}
# #1356: login resolution is now identity-aware, so the tea-branch fixture must
# offer the login the RUNNER's identity resolves to; otherwise every case below
# fails closed before reaching the branch under test.
#
# This does NOT make the suite hermetic, and it is not trying to. The API-path
# cases (5-7) need a usable Gitea token, and with an identity set the token path
# reads that seat's credential file rather than the GITEA_TOKEN exported above.
# So this suite passes only where the runner owns a real credential for its own
# identity, and fails with no identity at all -- on this branch and on its base
# alike. That is a pre-existing hole in the fixture, filed separately; pinning a
# synthetic identity here would only convert it into a confident-looking green.
_LOGIN_IDENT="${MOSAIC_GIT_IDENTITY:-}"
LOGIN_JSON='[{"name":"git.mosaicstack.dev","url":"https://git.mosaicstack.dev"}]'
if [[ -n "$_LOGIN_IDENT" ]]; then
LOGIN_JSON='[{"name":"mosaicstack-'"$_LOGIN_IDENT"'","url":"https://git.mosaicstack.dev"},{"name":"git.mosaicstack.dev","url":"https://git.mosaicstack.dev"}]'
fi
# The mocks must be the ones that run. Without this, a failed setup silently falls through
# to the real tea/curl and the "test" mutates the real provider.
@@ -10,12 +10,6 @@
set -euo pipefail
# HERMETICITY (#1356): this suite's subject is body quoting, not identity. An
# ambient MOSAIC_GIT_IDENTITY (every provisioned seat exports one) would make the
# identity ladder demand a per-seat login this fixture does not define, and the
# suite would fail for a reason it is not testing.
unset MOSAIC_GIT_IDENTITY
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/issue-create-body-safety}"
REPO_DIR="$WORK_DIR/repo"
@@ -67,18 +67,7 @@ cat > "$BIN_DIR/curl" <<'SH'
#!/usr/bin/env bash
set -euo pipefail
printf 'curl %s\n' "$*" >> "$MOSAIC_TEST_LOG"
# Repo roots: the pr-create API fallback resolves its base from the forge
# default_branch (T51-P2 WP5a). Exact-suffix so every other endpoint keeps
# the historical answer below.
url="${*: -1}"
case "$url" in
*/api/v1/repos/mosaicstack/stack)
printf '%s\n' '{"default_branch":"next"}'
;;
*)
printf '%s\n' '{"number":703}'
;;
esac
printf '%s\n' '{"number":703}'
SH
chmod +x "$BIN_DIR/tea" "$BIN_DIR/curl"
@@ -1,137 +0,0 @@
#!/usr/bin/env bash
# Regression: issue-view.sh must show comment BODIES, on both paths, and must name
# the failure tea actually reported instead of guessing a credential cause (#1357).
#
# Four defects, each with its own case below:
# F1 tea exits 1 in any repo with extensions.worktreeconfig=true; the wrapper must
# say so (git-config condition) and fall back to the API.
# F2 the API fallback dumped raw issue JSON, which carries only a comment COUNT.
# F3 the tea path never passed --comments, so tea prompted (non-interactively: nothing).
# F4 on ANY tea failure the wrapper printed the REVOKED OR STALE TOKEN note.
#
# Verification bar (plan §6): assert a real comment BODY appears, not a count and not
# `grep -c comment` (that instrument matched the issue title and read inverted).
#
# Hermetic: mock tea and curl on PATH, sandboxed repo. Resolves no real credentials.
set -euo pipefail
WORK_ROOT="${AGENT_WORK_ROOT:-${TMPDIR:-/tmp}}"
SANDBOX="$WORK_ROOT/issue-view-comments-test-$$"
MOCK_BIN="$SANDBOX/bin"; REPO_DIR="$SANDBOX/repo"; CALLS="$SANDBOX/calls.log"
cleanup() { rm -rf "$SANDBOX"; }
trap cleanup EXIT
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
TARGET="$SCRIPT_DIR/issue-view.sh"
[ -f "$TARGET" ] || { echo "FAIL: issue-view.sh not found beside this test"; exit 1; }
fail() { echo "FAIL: $*"; exit 1; }
mkdir -p "$MOCK_BIN" "$REPO_DIR" || fail "setup: cannot create sandbox under $WORK_ROOT"
: > "$CALLS" || fail "setup: cannot write calls log at $CALLS"
cd "$REPO_DIR" || fail "setup: cannot cd into $REPO_DIR"
git init -q || fail "setup: git init failed"
git remote add origin https://git.mosaicstack.dev/mosaicstack/stack.git || fail "setup: git remote add failed"
export PATH="$MOCK_BIN:$PATH" CALLS
export GITEA_URL="https://git.mosaicstack.dev"
export GITEA_TOKEN="redacted-test-token"
# The identity ladder must not reach for this seat's real login; the mock tea below
# defines the only login that exists in this sandbox.
unset MOSAIC_GIT_IDENTITY
# No fleet in the sandbox: on a host that runs one, get_gitea_token fails closed for an
# identity-less caller (by design), which would make this test measure the host, not
# the wrapper. An empty brain home makes the sandbox the same on every host.
export MOSAIC_BRAIN_HOME="$SANDBOX/brain"
mkdir -p "$MOSAIC_BRAIN_HOME" || fail "setup: cannot create sandbox brain home"
# Distinctive strings: a comment body that appears nowhere else, and an issue title
# that contains the word "comment" so a count-of-the-word instrument would misread.
BODY_MARKER="zebra-quill-comment-body-7731"
ISSUE_TITLE="wrapper never shows a comment"
# --- mock curl: serves the issue and its comments; logs every call --------------
cat > "$MOCK_BIN/curl" <<EOF
#!/bin/bash
url=""
while [ \$# -gt 0 ]; do
case "\$1" in
http*) url="\$1"; shift ;;
*) shift ;;
esac
done
printf 'curl %s\n' "\$url" >> "$CALLS"
case "\$url" in
*/issues/77/comments)
if [ "\${MOCK_NO_COMMENTS:-}" = "1" ]; then echo '[]'; else
echo '[{"id":1,"user":{"login":"alice"},"created_at":"2026-08-21T00:00:00Z","body":"$BODY_MARKER"}]'; fi ;;
*/issues/77)
echo '{"number":77,"title":"$ISSUE_TITLE","state":"open","user":{"login":"bob"},"created_at":"2026-08-21T00:00:00Z","labels":[],"milestone":null,"html_url":"https://git.mosaicstack.dev/mosaicstack/stack/issues/77","body":"issue body","comments":1}' ;;
*) echo '{}' ;;
esac
exit 0
EOF
chmod +x "$MOCK_BIN/curl"
# --- mock tea: MOCK_TEA_MODE selects the behaviour under test --------------------
# ok : prints the issue, and the comment body ONLY when --comments is passed (F3)
# wtconfig : exits 1 with the repositoryformatversion error (F1/F4)
# badtoken : exits 1 with tea's credential error (F4 control: credential wording allowed)
cat > "$MOCK_BIN/tea" <<EOF
#!/bin/bash
printf 'tea %s\n' "\$*" >> "$CALLS"
if [[ "\$*" == *"login list"* ]]; then
echo '[{"name":"git.mosaicstack.dev","url":"https://git.mosaicstack.dev"}]'; exit 0
fi
case "\${MOCK_TEA_MODE:-ok}" in
wtconfig) echo 'Error: core.repositoryformatversion does not support extension: worktreeconfig' >&2; exit 1 ;;
badtoken) echo 'Failed to create Gitea client: invalid username, password or token' >&2; exit 1 ;;
esac
echo "# #77 $ISSUE_TITLE (open)"
echo "issue body"
if [[ "\$*" == *"--comments"* ]]; then echo "$BODY_MARKER"; fi
exit 0
EOF
chmod +x "$MOCK_BIN/tea"
[ "$(command -v tea)" = "$MOCK_BIN/tea" ] || fail "setup: tea does not resolve inside the sandbox"
[ "$(command -v curl)" = "$MOCK_BIN/curl" ] || fail "setup: curl does not resolve inside the sandbox"
run() { bash "$TARGET" -i 77 >"$SANDBOX/out" 2>"$SANDBOX/err"; echo $?; }
# F3: tea path shows the comment body, which the mock emits only under --comments.
: > "$CALLS"
rc=$(MOCK_TEA_MODE=ok run)
[ "$rc" = 0 ] || fail "F3: expected rc=0 on the tea path, got $rc: $(cat "$SANDBOX/err")"
grep -q -- '--comments' "$CALLS" || fail "F3: tea was not invoked with --comments: $(cat "$CALLS")"
grep -q "$BODY_MARKER" "$SANDBOX/out" || fail "F3: comment body missing from tea-path output"
if grep -q '^curl' "$CALLS"; then fail "F3: tea path succeeded but the API fallback ran anyway"; fi
# F1 + F2: worktreeconfig failure is named as a git-config condition, falls back to
# the API, and the API rendering includes the comment BODY.
: > "$CALLS"
rc=$(MOCK_TEA_MODE=wtconfig run)
[ "$rc" = 0 ] || fail "F1: expected rc=0 via API fallback, got $rc: $(cat "$SANDBOX/err")"
grep -q 'worktreeconfig' "$SANDBOX/err" || fail "F1: stderr does not name the worktreeconfig cause: $(cat "$SANDBOX/err")"
grep -q 'not a credential problem' "$SANDBOX/err" || fail "F1: stderr does not rule out the credential cause"
grep -q 'issues/77/comments' "$CALLS" || fail "F2: API fallback never fetched /comments: $(cat "$CALLS")"
grep -q "$BODY_MARKER" "$SANDBOX/out" || fail "F2: comment body missing from API-path output"
grep -q "$ISSUE_TITLE" "$SANDBOX/out" || fail "F2: issue title missing from API-path output"
if grep -q 'REVOKED OR STALE' "$SANDBOX/err"; then fail "F4: stale-token note printed for a git-config failure"; fi
if grep -q '"comments": 1' "$SANDBOX/out"; then fail "F2: output is still raw JSON (comment count instead of bodies)"; fi
# F4 control: a real credential error from tea may still carry the credential note,
# and tea's own line must be relayed so the reader sees the actual cause.
: > "$CALLS"
rc=$(MOCK_TEA_MODE=badtoken run)
[ "$rc" = 0 ] || fail "F4 control: expected rc=0 via API fallback, got $rc"
grep -q 'invalid username, password or token' "$SANDBOX/err" || fail "F4: tea's own error line was not relayed"
if grep -q 'worktreeconfig' "$SANDBOX/err"; then fail "F4: git-config wording printed for a credential failure"; fi
# Negative control: an issue with no comments prints no comment section on the API
# path. Without this, a renderer that always prints a section would pass F2.
: > "$CALLS"
rc=$(MOCK_TEA_MODE=wtconfig MOCK_NO_COMMENTS=1 run)
[ "$rc" = 0 ] || fail "negative control: expected rc=0, got $rc"
if grep -q -- '--- Comments' "$SANDBOX/out"; then fail "negative control: comment section printed for an issue with no comments"; fi
if grep -q "$BODY_MARKER" "$SANDBOX/out"; then fail "negative control: a comment body appeared for an issue with no comments"; fi
echo "issue-view comments regression harness passed"
@@ -1,178 +0,0 @@
#!/usr/bin/env bash
# test-pr-create-fallback-default-base.sh — hermetic test for the API-fallback
# base resolution in pr-create.sh (T51-P2 WP5a / spec E4).
#
# The API-fallback payload historically hardcoded "base": "main", mistargeting
# every fallback PR on repos whose trunk is not main (e.g. mosaicstack/stack,
# default branch "next"). The fix: an explicit -B always wins; with none, the
# base is resolved from the provider API default_branch, and a failed
# resolution fails loud instead of guessing.
#
# Hermetic by construction: every curl invocation is a PATH-first stub; the
# fixture repo's remote is git.example.test (never dialed); HOME is a sandbox
# with no tea config (so the wrapper takes the API fallback path); GITEA_TOKEN
# comes from the environment. No real forge is contacted.
# shellcheck disable=SC2317
set -uo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/pr-create-fallback-base}"
PASS=0 FAIL=0 FAILED_CASES=""
ok() { PASS=$((PASS + 1)); }
bad() { FAIL=$((FAIL + 1)); FAILED_CASES="$FAILED_CASES $1"; printf 'FAIL: %s\n' "$1" >&2; }
assert_rc() { local d="$1" e="$2" a="$3"; [ "$e" = "$a" ] && ok || bad "$d (expected rc=$e got rc=$a)"; }
assert_eq() { local d="$1" e="$2" a="$3"; [ "$e" = "$a" ] && ok || bad "$d (expected [$e] got [$a])"; }
assert_contains() { local d="$1" h="$2" n="$3"; case "$h" in *"$n"*) ok ;; *) bad "$d (missing [$n])" ;; esac; }
json_field() { # $1 payload file, $2 field
python3 -c 'import json,sys; print(json.load(open(sys.argv[1])).get(sys.argv[2], ""))' "$1" "$2"
}
rm -rf "$WORK_DIR"
# ---- fixture -----------------------------------------------------------------
ROOT="$WORK_DIR/fixture"
TOOLS="$ROOT/tools/git"
mkdir -p "$TOOLS" "$ROOT/repo" "$ROOT/home" "$ROOT/stub"
cp "$SCRIPT_DIR/pr-create.sh" "$TOOLS/pr-create.sh"
cp "$SCRIPT_DIR/detect-platform.sh" "$TOOLS/detect-platform.sh"
git -C "$ROOT/repo" init -q -b fix/e4
git -C "$ROOT/repo" -c user.name=fixture -c user.email=fixture@test commit -q --allow-empty -m base
git -C "$ROOT/repo" remote add origin https://git.example.test/acme/widgets.git
# curl stub: GET repo -> default_branch JSON (or failure mode); POST pulls ->
# capture payload, answer with a minimal PR JSON. Every call is logged.
cat > "$ROOT/stub/curl" <<STUB
#!/usr/bin/env bash
set -u
mode="\${CURL_STUB_GET_MODE:-ok}"
printf '%s\n' "\$*" >> "$ROOT/curl-calls.log"
url="\${!#}"
if [[ "\$url" == */api/v1/repos/acme/widgets ]]; then
# repo GET (default-branch resolution). Modes cover the value shapes the
# resolver must accept or refuse (T51P2WP5AR B2/B3).
case "\$mode" in
ok) printf '%s\n' '{"id":1,"default_branch":"next","full_name":"acme/widgets"}' ;;
fail) echo "curl stub: simulated repo lookup failure" >&2; exit 1 ;;
fail-json) printf '%s\n' '{"default_branch":"next"}'; exit 22 ;;
null) printf '%s\n' '{"default_branch":null}' ;;
numeric) printf '%s\n' '{"default_branch":7}' ;;
blank) printf '%s\n' '{"default_branch":" "}' ;;
*) echo "curl stub: unknown GET mode \$mode" >&2; exit 1 ;;
esac
exit 0
fi
if [[ "\$url" == */api/v1/repos/acme/widgets/pulls ]]; then
# PR POST: capture the payload, emit a PR-shaped answer
while [[ \$# -gt 0 ]]; do
case "\$1" in
-d) printf '%s' "\$2" > "$ROOT/payload.json"; shift 2 ;;
*) shift ;;
esac
done
printf '%s\n' '{"number":42,"html_url":"https://git.example.test/acme/widgets/pulls/42"}'
exit 0
fi
echo "curl stub: unexpected URL \$url" >&2
exit 1
STUB
chmod +x "$ROOT/stub/curl"
run_pr_create() { # args... -> sets RC/OUT/ERR
RC=0
OUT=$(cd "$ROOT/repo" && env -i \
PATH="$ROOT/stub:/usr/bin:/bin" \
HOME="$ROOT/home" \
GITEA_TOKEN=stub-token \
bash "$TOOLS/pr-create.sh" "$@" 2>"$ROOT/err.txt")
RC=$?
ERR="$(cat "$ROOT/err.txt")"
}
calls_matching() { grep -c -- "$1" "$ROOT/curl-calls.log" 2>/dev/null || true; }
echo "== (1) no -B: fallback base resolves to the forge default branch, not main =="
: > "$ROOT/curl-calls.log"; rm -f "$ROOT/payload.json"
run_pr_create -t "fix thing"
assert_rc "rc" 0 "$RC"
assert_contains "API fallback path taken (tea login unresolvable in fixture)" "$ERR" "trying Gitea API fallback"
assert_eq "repo GET performed" 1 "$(calls_matching '/api/v1/repos/acme/widgets$')"
assert_eq "POST performed" 1 "$(calls_matching '/pulls$')"
assert_eq "payload base is forge default (next)" "next" "$(json_field "$ROOT/payload.json" base)"
assert_eq "payload head" "fix/e4" "$(json_field "$ROOT/payload.json" head)"
assert_eq "payload title" "fix thing" "$(json_field "$ROOT/payload.json" title)"
echo "== (2) explicit -B wins; the default branch is not consulted =="
: > "$ROOT/curl-calls.log"; rm -f "$ROOT/payload.json"
run_pr_create -t "fix thing" -B release/1.x
assert_rc "rc" 0 "$RC"
assert_eq "repo GET not consulted for explicit base" 0 "$(calls_matching '/api/v1/repos/acme/widgets$')"
assert_eq "payload base is the explicit -B" "release/1.x" "$(json_field "$ROOT/payload.json" base)"
echo "== (3) default-branch lookup failure: loud refusal, no POST =="
: > "$ROOT/curl-calls.log"; rm -f "$ROOT/payload.json"
RC=0
OUT=$(cd "$ROOT/repo" && env -i \
PATH="$ROOT/stub:/usr/bin:/bin" \
HOME="$ROOT/home" \
GITEA_TOKEN=stub-token \
CURL_STUB_GET_MODE=fail \
bash "$TOOLS/pr-create.sh" -t "fix thing" 2>"$ROOT/err.txt")
RC=$?
ERR="$(cat "$ROOT/err.txt")"
assert_rc "nonzero rc on unresolvable base" 1 "$RC"
assert_contains "loud error names -B" "$ERR" "could not resolve the forge default branch"
assert_contains "error names the remedy" "$ERR" "pass -B <branch> explicitly"
assert_eq "no POST issued" 0 "$(calls_matching '/pulls$')"
echo "== (4) payload never contains the literal fallback main =="
: > "$ROOT/curl-calls.log"; rm -f "$ROOT/payload.json"
run_pr_create -t "fix thing"
assert_rc "rc" 0 "$RC"
assert_eq "base field is next, never main" "next" "$(json_field "$ROOT/payload.json" base)"
echo "== (5) B2: HTTP failure with parseable JSON on stdout is a FAILED resolution =="
: > "$ROOT/curl-calls.log"; rm -f "$ROOT/payload.json"
RC=0
OUT=$(cd "$ROOT/repo" && env -i \
PATH="$ROOT/stub:/usr/bin:/bin" \
HOME="$ROOT/home" \
GITEA_TOKEN=stub-token \
CURL_STUB_GET_MODE=fail-json \
bash "$TOOLS/pr-create.sh" -t "fix thing" 2>"$ROOT/err.txt")
RC=$?
ERR="$(cat "$ROOT/err.txt")"
assert_rc "nonzero rc on HTTP failure despite valid JSON" 1 "$RC"
assert_contains "loud error names -B" "$ERR" "could not resolve the forge default branch"
assert_contains "error names the remedy" "$ERR" "pass -B <branch> explicitly"
assert_eq "no POST issued" 0 "$(calls_matching '/pulls$')"
echo "== (6) B3: null / numeric / blank default_branch are failed resolutions =="
for bad in null numeric blank; do
: > "$ROOT/curl-calls.log"; rm -f "$ROOT/payload.json"
RC=0
OUT=$(cd "$ROOT/repo" && env -i \
PATH="$ROOT/stub:/usr/bin:/bin" \
HOME="$ROOT/home" \
GITEA_TOKEN=stub-token \
CURL_STUB_GET_MODE="$bad" \
bash "$TOOLS/pr-create.sh" -t "fix thing" 2>"$ROOT/err.txt")
RC=$?
ERR="$(cat "$ROOT/err.txt")"
assert_rc "B3 $bad: nonzero rc" 1 "$RC"
assert_contains "B3 $bad: loud error" "$ERR" "could not resolve the forge default branch"
assert_eq "B3 $bad: no POST issued" 0 "$(calls_matching '/pulls$')"
done
echo
echo "pass=$PASS fail=$FAIL"
if [ "$FAIL" -gt 0 ]; then
echo "FAILED CASES:$FAILED_CASES"
exit 1
fi
echo "ALL GREEN"
@@ -1,84 +0,0 @@
#!/usr/bin/env bash
# pr-merge must forward --no-ci-expected to the queue guard, and only then.
# The flag is the sanctioned merge path for a repository with no CI configured
# (see test-ci-queue-wait-no-ci-expected.sh for the guard-side semantics);
# this harness pins only the pass-through: present when requested, absent when
# not, with the rest of the guard invocation unchanged.
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
WORK_DIR="${MOSAIC_TEST_WORK_DIR:-$PWD/.mosaic-test-work/pr-merge-no-ci-expected}"
FIXTURE_DIR="$WORK_DIR/tools/git"
CALL_LOG="$WORK_DIR/queue-call.log"
rm -rf "$WORK_DIR"
mkdir -p "$FIXTURE_DIR"
cp "$SCRIPT_DIR/pr-merge.sh" "$FIXTURE_DIR/pr-merge.sh"
cp "$SCRIPT_DIR/detect-platform.sh" "$FIXTURE_DIR/detect-platform.sh"
cat > "$FIXTURE_DIR/pr-metadata.sh" <<'SH'
#!/usr/bin/env bash
printf '%s\n' '{"baseRefName":"main","baseRepository":"mosaicstack/stack","headRefName":"fix/no-ci-fixture","headRefOid":"0123456789abcdef0123456789abcdef01234567","headRepository":"mosaicstack/stack"}'
SH
cat > "$FIXTURE_DIR/ci-queue-wait.sh" <<'SH'
#!/usr/bin/env bash
printf '%s\n' "$*" > "${MOSAIC_QUEUE_CALL_LOG:?}"
exit 42
SH
chmod +x "$FIXTURE_DIR"/*.sh
run_merge() {
(
cd "$WORK_DIR"
export MOSAIC_QUEUE_CALL_LOG="$CALL_LOG"
"$FIXTURE_DIR/pr-merge.sh" -n 123 "$@"
) >/dev/null 2>&1
}
fail=0
# With the flag: it must reach the guard invocation.
: > "$CALL_LOG"
set +e
run_merge --no-ci-expected
rc_with=$?
set -e
if [[ "$rc_with" -ne 42 ]]; then
echo "FAIL(with): expected queue stub rc=42 to propagate, got $rc_with" >&2
fail=1
elif ! grep -q -- '--no-ci-expected' "$CALL_LOG"; then
echo "FAIL(with): --no-ci-expected did not reach the queue guard" >&2
cat "$CALL_LOG" >&2
fail=1
fi
# The rest of the guard invocation is unchanged by the flag.
for required in '--purpose merge' '-B fix/no-ci-fixture' '-R mosaicstack/stack' \
'--sha 0123456789abcdef0123456789abcdef01234567'; do
if ! grep -qF -- "$required" "$CALL_LOG"; then
echo "FAIL(with): guard invocation lost '$required'" >&2
cat "$CALL_LOG" >&2
fail=1
fi
done
# Without the flag: it must NOT appear in the guard invocation.
: > "$CALL_LOG"
set +e
run_merge
rc_without=$?
set -e
if [[ "$rc_without" -ne 42 ]]; then
echo "FAIL(without): expected queue stub rc=42 to propagate, got $rc_without" >&2
fail=1
elif grep -q -- '--no-ci-expected' "$CALL_LOG"; then
echo "FAIL(without): --no-ci-expected reached the guard without being requested" >&2
cat "$CALL_LOG" >&2
fail=1
fi
if [[ "$fail" -eq 0 ]]; then
echo "pr-merge no-ci-expected pass-through regression passed"
fi
exit "$fail"
@@ -13,6 +13,7 @@
# --- tools/git: the #1007 five — non-hermetic, resolve real credentials ---
packages/mosaic/framework/tools/git/test-pr-merge-gitea-empty-uid.sh | resolves real credentials (#1007 census); joins CI after the wrapper-half hermeticity fix (git -C scoping)
packages/mosaic/framework/tools/git/test-issue-create-interactive-auth.sh | resolves real credentials (#1007 census); joins CI after the wrapper-half hermeticity fix
packages/mosaic/framework/tools/git/test-gitea-login-resolution.sh | resolves real credentials (#1007 census); joins CI after the wrapper-half hermeticity fix
packages/mosaic/framework/tools/git/test-pr-metadata-gitea.sh | resolves real credentials (#1007 census, fourth entry via family-grep); joins CI after the wrapper-half hermeticity fix
packages/mosaic/framework/tools/git/test-issue-comment-readback.sh | resolves real credentials (#1007 census, fifth entry); joins CI after the wrapper-half hermeticity fix
@@ -1,289 +0,0 @@
#!/usr/bin/env bash
# test-validate-repo-json.sh — hostile-input suite for the T51 declaration validator.
# (Vendored with validate-repo-json.sh from mosaic-brain @ 515bcbab — see the
# validator header for provenance.)
#
# Hermetic: all fixtures in a tracked mktemp sandbox removed by an EXIT trap
# (pass and fail paths both — zero residue). No network, no real repos, no host
# state mutated. MOSAIC_HOST_ROOT is set/unset per arm via env only.
#
# T51P2RW1: arms extended per review T51P2R1 (F1-F5): root gate for ordinary
# v2 declarations (unset AND explicitly empty; display warns), git-grammar
# branch arms (double slash, dot component, control byte), contract-escape
# arms (list enums, invalid UTF-8, NaN — one VALIDATION_ERROR line, never a
# traceback), remote normalization (.git/ ordering, port preservation), and
# mirror-component fullmatch arms (trailing newline, control bytes).
set -u
HERE=$(cd "$(dirname "$0")" && pwd)
V="$HERE/validate-repo-json.sh"
PASS=0; FAIL=0; FAILED=""
ok() { PASS=$((PASS+1)); }
bad() { FAIL=$((FAIL+1)); FAILED="$FAILED $1"; printf 'FAIL: %s\n' "$1" >&2; }
SB=$(mktemp -d "${TMPDIR:-/tmp}/vrj-test.XXXXXX")
trap 'rm -rf "$SB"' EXIT
fx() { printf '%s' "$2" > "$SB/$1"; }
run() { # [env KV=V ...] -- args...
local envs=()
while [ "$1" != "--" ]; do envs+=("$1"); shift; done; shift
OUT=$(env "${envs[@]:-_=_}" bash "$V" "$@" 2>&1 < /dev/null; echo "__RC__$?")
RC=${OUT##*__RC__}; OUT=${OUT%__RC__*}; OUT=${OUT%$'\n'}
}
expect_ok() { local d="$1"; shift; run "$@"; if [ "$RC" = 0 ] && printf '%s' "$OUT" | grep -q '^OK'; then ok; else bad "$d (rc=$RC out=$(printf '%s' "$OUT" | head -1))"; fi; }
expect_err() { # desc expected-substring [env... -- args...]
local d="$1" sub="$2"; shift 2
run "$@"
if [ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "VALIDATION_ERROR.*$sub"; then ok
else bad "$d (rc=$RC, wanted error ~$sub, got: ${OUT%%$'\n'*})"; fi
}
expect_err_notrace() { # like expect_err, plus no traceback anywhere in output
local d="$1" sub="$2"; shift 2
run "$@"
if [ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "VALIDATION_ERROR.*$sub" && ! printf '%s' "$OUT" | grep -q "Traceback"; then ok
else bad "$d (rc=$RC, wanted clean error ~$sub, got: ${OUT%%$'\n'*})"; fi
}
STACK='{"schema_version":2,"integration_trunk":"next","release_branch":"main","flow":"trunk-release","canonical_remote":"https://git.mosaicstack.dev/mosaicstack/stack","canonical_clone":"host:/src/mosaic-stack","worktree_root":"host:/src/mosaic-stack-worktrees","worktree_policy":"orchestrator-precreated","notes":"x"}'
BRAIN='{"schema_version":2,"integration_trunk":"main","release_branch":"main","flow":"direct","canonical_remote":"https://git.example.invalid/acme/brain","canonical_clone":"host:/.mosaic","worktree_root":"host:/.mosaic-worktrees","worktree_policy":"orchestrator-precreated"}'
ROOT="$SB/hostroot"; mkdir -p "$ROOT"
echo "== (0) syntax + version =="
bash -n "$V" && ok || bad "bash -n"
run -- --version; [ "$RC" = 0 ] && case "$OUT" in validate-repo-json\ *) ok ;; *) bad "version output" ;; esac || bad "version rc"
echo "== (1) spec examples: stack + brain OK (root set) =="
fx stack.json "$STACK"; fx brain.json "$BRAIN"
expect_ok a1 MOSAIC_HOST_ROOT=$ROOT -- "$SB/stack.json"
expect_ok a2 MOSAIC_HOST_ROOT=$ROOT -- "$SB/brain.json"
echo "== (2) malformed JSON (stable contract, no traceback) =="
fx bad.json '{"schema_version": 2, '
expect_err_notrace b1 "json:" -- "$SB/bad.json"
fx arr.json '[1,2]'
expect_err_notrace b2 "top level" -- "$SB/arr.json"
printf '\xff\xfe{"schema_version":2}' > "$SB/utf8.json"
expect_err_notrace b3 "UTF-8" MOSAIC_HOST_ROOT=$ROOT -- "$SB/utf8.json"
fx nan.json '{"schema_version":NaN}'
expect_err_notrace b4 "malformed JSON" MOSAIC_HOST_ROOT=$ROOT -- "$SB/nan.json"
echo "== (3) unknown schema_version = ABSENT-loud =="
fx v3.json "${STACK/schema_version\":2/schema_version\":3}"
expect_err c1 "schema_version" MOSAIC_HOST_ROOT=$ROOT -- "$SB/v3.json"
echo "== (4) v1 mode + authoring rule (v1 consumes no paths: no root needed) =="
fx v1.json '{"integration_trunk":"next","release_branch":"main"}'
expect_ok d1 -- "$SB/v1.json"
expect_err d2 "schema_version" -- --require-v2 "$SB/v1.json"
fx v1x.json '{"integration_trunk":"next","release_branch":"main","notes":"no"}'
expect_err d3 "x_extensions" -- "$SB/v1x.json"
echo "== (5) unknown top-level key rejected; x_extensions home OK =="
fx unk.json "${STACK%\}*},\"typo_key\":1}"
expect_err e1 "typo_key" MOSAIC_HOST_ROOT=$ROOT -- "$SB/unk.json"
fx ext.json "${STACK%\}*},\"x_extensions\":{\"future\":true}}"
expect_ok e2 MOSAIC_HOST_ROOT=$ROOT -- "$SB/ext.json"
echo "== (6) flow: required (no defaulting) + cross-field =="
fx noflow.json "$(printf '%s' "$STACK" | python3 -c 'import json,sys; d=json.load(sys.stdin); del d["flow"]; print(json.dumps(d))')"
expect_err f1 "flow" MOSAIC_HOST_ROOT=$ROOT -- "$SB/noflow.json"
fx xdirect.json "${STACK/\"trunk-release\"/\"direct\"}"
expect_err f2 "direct" MOSAIC_HOST_ROOT=$ROOT -- "$SB/xdirect.json"
fx xtr.json "${BRAIN/\"direct\"/\"trunk-release\"}"
expect_err f3 "trunk-release" MOSAIC_HOST_ROOT=$ROOT -- "$SB/xtr.json"
echo "== (7) dot-segment / empty-segment / tilde escapes =="
fx dots.json "${STACK/host:\/src\/mosaic-stack\"/host:/src/../secrets\"}"
expect_err g1 "dot segment" MOSAIC_HOST_ROOT=$ROOT -- "$SB/dots.json"
fx dot1.json "${STACK/host:\/src\/mosaic-stack\"/host:/src/./mosaic-stack\"}"
expect_err g2 "dot segment" MOSAIC_HOST_ROOT=$ROOT -- "$SB/dot1.json"
fx empty.json "${STACK/host:\/src\/mosaic-stack\"/host://src/mosaic-stack\"}"
expect_err g3 "empty segment" MOSAIC_HOST_ROOT=$ROOT -- "$SB/empty.json"
fx tild.json "${STACK/host:\/src\/mosaic-stack\"/~jw/src/mosaic-stack\"}"
expect_err g4 "tilde" MOSAIC_HOST_ROOT=$ROOT -- "$SB/tild.json"
fx tailslash.json "${STACK/host:\/src\/mosaic-stack\"/host:/src/mosaic-stack/\"}"
expect_err g5 "empty segment" MOSAIC_HOST_ROOT=$ROOT -- "$SB/tailslash.json"
fx noanchor.json "${STACK/host:\/src\/mosaic-stack\"//src/mosaic-stack\"}"
expect_err g6 "host:/" MOSAIC_HOST_ROOT=$ROOT -- "$SB/noanchor.json"
echo "== (8) branch-name grammar (delegated to git check-ref-format, F2) =="
fx badbr.json "${STACK/\"next\"/\"bad..name\"}"
expect_err h1 "branch name" MOSAIC_HOST_ROOT=$ROOT -- "$SB/badbr.json"
fx sp.json "${STACK/\"next\"/\"fea ture\"}"
expect_err h2 "branch name" MOSAIC_HOST_ROOT=$ROOT -- "$SB/sp.json"
fx lock.json "${STACK/\"next\"/\"feature/x.lock\"}"
expect_err h3 "branch name" MOSAIC_HOST_ROOT=$ROOT -- "$SB/lock.json"
fx slash.json "${STACK/\"next\"/\"feature/x\"}"
expect_ok h4 MOSAIC_HOST_ROOT=$ROOT -- "$SB/slash.json"
fx dslash.json "${STACK/\"next\"/\"feature//x\"}"
expect_err h5 "branch name" MOSAIC_HOST_ROOT=$ROOT -- "$SB/dslash.json"
fx hidden.json "${STACK/\"next\"/\"feature/.hidden\"}"
expect_err h6 "branch name" MOSAIC_HOST_ROOT=$ROOT -- "$SB/hidden.json"
fx ctrl.json "$(printf '%s' "$STACK" | python3 -c 'import json,sys; d=json.load(sys.stdin); d["integration_trunk"]="feature/\x01x"; print(json.dumps(d))')"
expect_err h7 "branch name" MOSAIC_HOST_ROOT=$ROOT -- "$SB/ctrl.json"
echo "== (8b) reflog shorthand rejected independent of ambient checkout history (B1) =="
# Hermetic repo WITH checkout history: proves '@{-1}' (which git would expand to
# 'main' from THIS repo's reflog) is still refused by the pre-delegation gate.
HISTREPO="$SB/histrepo"; mkdir -p "$HISTREPO"
(cd "$HISTREPO" && git init -q -b main . \
&& git -c user.name=t -c user.email=t@t commit -q --allow-empty -m m \
&& git checkout -q -b feature/x \
&& git checkout -q main \
&& git check-ref-format --branch "@{-1}" >/dev/null 2>&1 && echo "ambient-expandable" || echo "not-expandable") \
| grep -q ambient-expandable && ok || bad "fixture repo failed to make @{-1} expandable"
fx atminus1.json "$(printf '%s' "$STACK" | python3 -c 'import json,sys; d=json.load(sys.stdin); d["integration_trunk"]="@{-1}"; print(json.dumps(d))')"
# Run the validator from INSIDE the history repo via command substitution so the
# assertion runs in the PARENT shell (T51P2R3 B1: the previous ( subshell ) form
# mutated ok/bad counters only in a dead subshell — FAIL printed, suite rc 0).
OUTX=$(cd "$HISTREPO" && MOSAIC_HOST_ROOT=$ROOT bash "$V" "$SB/atminus1.json" 2>&1 </dev/null; echo "__RC__$?")
RCX=${OUTX##*__RC__}
if [ "$RCX" = 1 ] && printf '%s' "$OUTX" | grep -q "VALIDATION_ERROR.*@{"; then ok
else bad "@{-1} must be rejected inside a repo with checkout history (got rc=$RCX)"; fi
fx atbrace.json "$(printf '%s' "$STACK" | python3 -c 'import json,sys; d=json.load(sys.stdin); d["integration_trunk"]="@{u}"; print(json.dumps(d))')"
expect_err h9 "@{" MOSAIC_HOST_ROOT=$ROOT -- "$SB/atbrace.json"
echo "== (9) canonical_remote: userinfo, list-type, normalization (F3/F4) =="
fx user.json "${STACK/https:\/\/git.mosaicstack.dev/https:\/\/bot:s3cret@git.mosaicstack.dev}"
expect_err_notrace i1 "userinfo" MOSAIC_HOST_ROOT=$ROOT -- "$SB/user.json"
fx listflow.json "${STACK/\"trunk-release\"/[\"trunk-release\"]}"
expect_err_notrace i2 "flow" MOSAIC_HOST_ROOT=$ROOT -- "$SB/listflow.json"
fx listpol.json "${STACK/\"orchestrator-precreated\"/[\"tool-managed\"]}"
expect_err_notrace i3 "worktree_policy" MOSAIC_HOST_ROOT=$ROOT -- "$SB/listpol.json"
run -- --normalize-remote "HTTPS://Git.Example.Invalid/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://git.example.invalid/o/r" ] && ok || bad "norm .git/case ($OUT)"
run -- --normalize-remote "https://git.mosaicstack.dev/mosaicstack/stack/"
[ "$RC" = 0 ] && [ "$OUT" = "https://git.mosaicstack.dev/mosaicstack/stack" ] && ok || bad "norm trailing slash ($OUT)"
run -- --normalize-remote "git.mosaicstack.dev/mosaicstack/stack"
[ "$RC" = 1 ] && ok || bad "schemeless must fail"
run -- --normalize-remote "HTTPS://Git.Example.Invalid/o/r.git/"
[ "$RC" = 0 ] && [ "$OUT" = "https://git.example.invalid/o/r" ] && ok || bad "norm .git-then-slash ($OUT)"
run -- --normalize-remote "https://Git.Example.Invalid:8443/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://git.example.invalid:8443/o/r" ] && ok || bad "port must be preserved ($OUT)"
run -- --normalize-remote "https://[2001:db8::1]:8443/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://[2001:db8::1]:8443/o/r" ] && ok || bad "IPv6 must stay bracketed with port ($OUT)"
run -- --normalize-remote "https://[2001:db8::1]/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://[2001:db8::1]/o/r" ] && ok || bad "IPv6 must stay bracketed ($OUT)"
run -- --normalize-remote "https://Git.Example.Invalid:0/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://git.example.invalid:0/o/r" ] && ok || bad "explicit port 0 must be preserved ($OUT)"
run -- --normalize-remote "https://[::1].evil.example/o/r.git"
[ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "canonical_remote" && ok || bad "suffix after ] must be rejected (.evil.example)"
run -- --normalize-remote "https://[::1]x:8443/o/r.git"
[ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "canonical_remote" && ok || bad "suffix after ] must be rejected (x:8443)"
run -- --normalize-remote "https://[::1]x/o/r.git"
[ "$RC" = 1 ] && ok || bad "suffix after ] must be rejected (x)"
echo "== (9b) IPvFuture bracketed authorities (R4-B1: guard keys off raw netloc) =="
run -- --normalize-remote "https://[v1.fe80]/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://[v1.fe80]/o/r" ] && ok || bad "valid IPvFuture must keep brackets ($OUT)"
run -- --normalize-remote "https://[vF.foo]:8443/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://[vf.foo]:8443/o/r" ] && ok || bad "valid IPvFuture+port must keep brackets ($OUT)"
run -- --normalize-remote "https://[v1.fe80]evil/o/r.git"
[ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "canonical_remote" && ok || bad "IPvFuture suffix must be rejected (evil)"
run -- --normalize-remote "https://[v1.fe80].evil.example/o/r.git"
[ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "canonical_remote" && ok || bad "IPvFuture suffix must be rejected (.evil.example)"
run -- --normalize-remote "https://[vF.foo]x:8443/o/r.git"
[ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "canonical_remote" && ok || bad "IPvFuture suffix must be rejected (x:8443)"
echo "== (9d) non-bracketed authority grammar (R6-B1) =="
for U in "https://:8443/o/r.git" "https://bad host/o/r.git" "https://bad^host/o/r.git" "https://bad\\host/o/r.git" "https://bad%zz/o/r.git" "https://bad%2/o/r.git" "https://bad%/o/r.git"; do
run -- --normalize-remote "$U"
if [ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "VALIDATION_ERROR canonical_remote"; then ok
else bad "non-bracketed authority must be rejected: $U (rc=$RC out=$OUT)"; fi
done
run -- --normalize-remote "https://git.mosaicstack.dev:9000/mosaicstack/stack"
[ "$RC" = 0 ] && [ "$OUT" = "https://git.mosaicstack.dev:9000/mosaicstack/stack" ] && ok || bad "valid host:port unchanged ($OUT)"
run -- --normalize-remote "https://192.168.1.10:8443/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://192.168.1.10:8443/o/r" ] && ok || bad "IPv4 reg-name stays valid ($OUT)"
run -- --normalize-remote "https://bad%2Fx/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://bad%2fx/o/r" ] && ok || bad "complete %HH must stay legal, case-normalized ($OUT)"
echo "== (9e) ASCII-only authority bytes (R7-B1) =="
# isolated port arm (R8): VALID ASCII host + full-width-digit port ONLY —
# unconfounded, so restoring Unicode-aware isdigit() goes red right here.
run -- --normalize-remote "https://git.example.invalid:443/o/r.git"
if [ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "ASCII digits only"; then ok
else bad "full-width-digit port on a VALID host must be rejected with the port reason (rc=$RC out=$OUT)"; fi
for U in "https://éxample.invalid/o/r.git" "https://例え.テスト/o/r.git" "https://fullwidth.invalid/o/r.git" "https://mosaicstack.dev:443/o/r.git"; do
run -- --normalize-remote "$U"
if [ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "VALIDATION_ERROR canonical_remote"; then ok
else bad "non-ASCII authority must be rejected: $U (rc=$RC out=$OUT)"; fi
done
run -- --normalize-remote "https://xn--xample-9ua.invalid/o/r.git"
[ "$RC" = 0 ] && [ "$OUT" = "https://xn--xample-9ua.invalid/o/r" ] && ok || bad "punycode xn-- host must stay legal ($OUT)"
echo "== (9c) bracket-payload grammar + raw control bytes (R5-B1) =="
for P in "v1. " "v1.a b" "v1.a^b" "v1.a\\b" "v1.%20" "not-an-ip" "::gg::1"; do
run -- --normalize-remote "https://[$P]/o/r.git"
if [ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "VALIDATION_ERROR canonical_remote"; then ok
else bad "bracket payload [$P] must be rejected (rc=$RC out=$OUT)"; fi
done
for CB in $'\t' $'\n' $'\r'; do
run -- --normalize-remote "https://[v1.a${CB}b]/o/r.git"
if [ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "control byte"; then ok
else bad "raw control byte must be rejected before urlsplit (rc=$RC out=$OUT)"; fi
done
run -- --normalize-remote "https://[v1.fe80%zone]/o/r.git"
[ "$RC" = 1 ] && ok || bad "percent (not in RFC host grammar) must be rejected ($OUT)"
run -- --normalize-remote "https://[fe80::1%eth0]/o/r.git"
[ "$RC" = 1 ] && printf '%s' "$OUT" | grep -q "canonical_remote" && ok || bad "IPv6 zone-id (not RFC host grammar) must be rejected ($OUT)"
echo "== (10) root gate: every v2 managed validation fails closed (F1) =="
# NOTE (spec §4.5): host:/ paths resolve UNDER MOSAIC_HOST_ROOT by construction,
# so tool-managed is not declarable today — the cross-check fails for every
# host:/ root until the anchor scheme grows an outside-root form (J3 era).
expect_err j1 "MOSAIC_HOST_ROOT" MOSAIC_HOST_ROOT= -- "$SB/stack.json"
expect_err j2 "MOSAIC_HOST_ROOT" MOSAIC_HOST_ROOT= -- "$SB/brain.json"
expect_err j3 "MOSAIC_HOST_ROOT" MOSAIC_HOST_ROOT= -- "$SB/stack.json"
# rider (T51P2R2): genuine ABSENCE, not just explicitly empty — captured via
# command substitution, asserted in the parent shell (no subshell-counter shape).
OUTU=$(cd "$SB" && env -u MOSAIC_HOST_ROOT bash "$V" "$SB/stack.json" 2>&1 </dev/null; echo "__RC__$?")
RCU=${OUTU##*__RC__}
if [ "$RCU" = 1 ] && printf '%s' "$OUTU" | grep -q "VALIDATION_ERROR.*MOSAIC_HOST_ROOT"; then ok
else bad "unset-by-absence root must fail closed in managed mode (rc=$RCU)"; fi
run MOSAIC_HOST_ROOT= -- --mode display "$SB/stack.json"
if [ "$RC" = 0 ] && printf '%s' "$OUT" | grep -q '^OK' && printf '%s' "$OUT" | grep -q "host root unset"; then ok
else bad "display-mode unset must pass with the specified warning (rc=$RC)"; fi
run MOSAIC_HOST_ROOT= -- --mode display "$SB/brain.json"
if [ "$RC" = 0 ] && printf '%s' "$OUT" | grep -q "host root unset"; then ok
else bad "display-mode warn missing for brain fixture"; fi
TM='{"schema_version":2,"integration_trunk":"next","release_branch":"main","flow":"trunk-release","canonical_remote":"https://git.mosaicstack.dev/mosaicstack/stack","canonical_clone":"host:/src/mosaic-stack","worktree_root":"host:/src/mosaic-stack-worktrees","worktree_policy":"tool-managed"}'
fx tm.json "$TM"; mkdir -p "$ROOT/src"
expect_err j4 "inside MOSAIC_HOST_ROOT" MOSAIC_HOST_ROOT=$ROOT -- "$SB/tm.json"
fx tm_noroot.json "$(printf '%s' "$TM" | python3 -c 'import json,sys; d=json.load(sys.stdin); del d["worktree_root"]; print(json.dumps(d))')"
expect_err j5 "worktree_root" MOSAIC_HOST_ROOT=$ROOT -- "$SB/tm_noroot.json"
echo "== (10b) symlink escape cannot fake outside-ness (B3 fix: lexical containment) =="
OUTSIDE="$SB/outside-target"; mkdir -p "$OUTSIDE"
ln -s "$OUTSIDE" "$ROOT/escape"
TM_ESC="${TM/host:\/src\/mosaic-stack-worktrees/host:/escape/worktrees}"
fx tmsym.json "$TM_ESC"
expect_err j6 "inside MOSAIC_HOST_ROOT" MOSAIC_HOST_ROOT=$ROOT -- "$SB/tmsym.json"
echo "== (11) mirror-path components: collision/delimiter/control fixtures (F5) =="
run -- --mirror-path git.mosaicstack.dev mosaicstack stack
[ "$RC" = 0 ] && [ "$OUT" = "projects/git.mosaicstack.dev/mosaicstack/stack/repo.json" ] && ok || bad "mirror path ok ($OUT)"
expect_err k1 "mirror-component" -- --mirror-path "git.mosaicstack.dev" "a__b" "c"
expect_err k2 "mirror-component" -- --mirror-path "git.mosaicstack.dev" "a" "b__c"
expect_err k3 "mirror-component" -- --mirror-path "git.mosaicstack.dev/x" "a" "b"
expect_err k4 "mirror-component" -- --mirror-path "git.mosaicstack.dev" "A" "B"
expect_err k5 "mirror-component" -- --mirror-path "git.mosaicstack.dev" "" "stack"
run -- --mirror-path git.mosaicstack.dev a b.c
[ "$RC" = 0 ] && [ "$OUT" = "projects/git.mosaicstack.dev/a/b.c/repo.json" ] && ok || bad "distinct path ($OUT)"
run -- --mirror-path $'git.example.invalid\n' owner repo
[ "$RC" = 1 ] && ok || bad "trailing-newline host must be rejected (fullmatch)"
run -- --mirror-path $'git.\texample' owner repo
[ "$RC" = 1 ] && ok || bad "control-byte host must be rejected"
echo "== (12) missing required keys =="
for key in release_branch canonical_clone; do
fx miss.json "$(printf '%s' "$STACK" | python3 -c "import json,sys; d=json.load(sys.stdin); del d['$key']; print(json.dumps(d))")"
expect_err "l-$key" "$key" MOSAIC_HOST_ROOT=$ROOT -- "$SB/miss.json"
done
echo "== (13) absent file =="
expect_err m1 "file" -- "$SB/nonexistent.json"
echo
echo "pass=$PASS fail=$FAIL"
if [ "$FAIL" -gt 0 ]; then echo "FAILED:$FAILED"; exit 1; fi
echo "ALL GREEN"
@@ -1,386 +0,0 @@
#!/usr/bin/env bash
# validate-repo-json.sh — declaration validator for T51 repo structure declarations.
#
# Spec of record: docs/plans/2026-08-23_repo-structure-declaration.md @ 1896adc1
# (R3). Implements the spec's validation surface: schema v1/v2 (§1.2), host:/
# path grammar with canonical segment normalization — empty/./.. rejected
# BEFORE resolution — and tilde rejection (§1.2a), mirror path component
# validation (§3.1), cross-field rules (§5.2), remote normalization (§5.3).
#
# PROVENANCE (T51 WP5c vendoring): ported verbatim from the mosaic-brain tree —
# tools/repo-structure-decl/validate-repo-json.sh @ brain main merge 515bcbab
# (PR 28, wave-1 R9 PASS, 101-arm suite green). This file is now the framework
# home per spec §5.1 ("shipped in the framework package"); the brain copy is the
# development origin. Re-sync rule: changes land here via reviewed PR and are
# back-ported to the brain tree (or the brain copy retires) — never fork silently.
# Validator version at port: 1.1.0+t51spec-r3+t51p2rw1 (R2-R8 rework included).
# No operator literal appears in this file; the host root is read from
# MOSAIC_HOST_ROOT configuration only.
#
# Unset-root semantics (spec §1.2a, fail-closed; T51P2R1 F1): v2 declarations
# always consume a path (canonical_clone is required), so in --mode managed an
# unset OR EMPTY MOSAIC_HOST_ROOT is a VALIDATION_ERROR for every v2 file —
# not only tool-managed. In --mode display it warns and omits root-dependent
# resolution; grammar checks still run.
#
# Error contract (T51P2R1 F3): every malformed input — bad UTF-8, non-RFC JSON
# constants (NaN/Infinity), wrong-typed enums, anything unexpected — yields
# exactly one stable VALIDATION_ERROR line and exit 1. No traceback ever
# escapes. Branch names are validated by delegating to `git check-ref-format
# --branch` (F2), translated into this contract.
#
# Usage:
# validate-repo-json.sh <repo.json> [--mode managed|display] [--require-v2]
# validate-repo-json.sh --mirror-path <host> <owner> <repo> # §3.1 component check
# validate-repo-json.sh --normalize-remote <url> # §5.3, prints normalized
# validate-repo-json.sh --version
#
# Output: OK (exit 0) | VALIDATION_ERROR <key>: <reason> (exit 1) | warnings on stderr.
set -euo pipefail
VERSION="1.1.0+t51spec-r3+t51p2rw1"
if [ "${1:-}" = "--version" ]; then echo "validate-repo-json $VERSION"; exit 0; fi
exec python3 - "$@" <<'PYEOF'
import json, os, re, subprocess, sys, urllib.parse
def err(key, reason):
print(f"VALIDATION_ERROR {key}: {reason}")
sys.exit(1)
def warn(msg):
print(f"warning: {msg}", file=sys.stderr)
def main():
ARGS = sys.argv[1:]
MODE = "managed"
REQUIRE_V2 = False
# ---- subcommands first (they take no file argument) ----
def check_mirror_components(host, owner, repo):
# fullmatch: '$' must bind at true end (F5 — a trailing newline must
# NOT pass); the charset excludes control bytes outright.
comp_re = re.compile(r"[a-z0-9][a-z0-9.-]*")
for label, value in (("host", host), ("owner", owner), ("repo", repo)):
if not isinstance(value, str) or not value or not comp_re.fullmatch(value):
err("mirror-component", f"{label} {value!r} fails §3.1 charset ^[a-z0-9][a-z0-9.-]*$ (fullmatch, no '/', no delimiter, no control bytes)")
return f"projects/{host}/{owner}/{repo}/repo.json"
def normalize_remote(url):
# R5-B1 part 1: reject raw control bytes BEFORE urlsplit — urlsplit
# silently strips TAB/LF/CR, so different input bytes would normalize
# to a different host. The raw bytes ARE the input; nothing may rewrite them.
for ch in url:
if ord(ch) < 0x20 or ord(ch) == 0x7F:
err("canonical_remote", "control byte in URL rejected before parsing (urlsplit would strip it and change the host)")
try:
p = urllib.parse.urlsplit(url)
except ValueError as e:
# py3.12 urlsplit itself validates bracketed hosts (ipaddress) and
# raises for garbage authorities — translate, never traceback.
err("canonical_remote", f"invalid URL authority: {e}")
if not p.scheme or not p.netloc:
err("canonical_remote", f"not a URL with scheme+host: {url!r}")
if p.username or p.password or "@" in (p.netloc or ""):
err("canonical_remote", "userinfo in URL is rejected (§5.3)")
scheme = p.scheme.lower()
# T51P2R4 B1: bracketing is detected from the RAW netloc ('[' prefix),
# not from a ':' in the parsed hostname — IPvFuture literals ([v1.fe80])
# contain no colon and must not bypass the raw-authority proof.
bracketed = p.netloc.startswith("[")
if bracketed:
# Prove the RAW authority is exactly '[host]' + optional ':port'
# (case-normalized); any text after ']' is hostile/truncated input,
# rejected — never silently rewritten.
import re as _re
import ipaddress as _ip
m = _re.fullmatch(r"\[([^\]]*)\](?::([0-9]+))?", p.netloc)
if not m:
err("canonical_remote", f"malformed bracketed authority {p.netloc!r}: text after ']' is rejected (no silent truncation)")
payload = m.group(1)
# R5-B1 part 2: the bracket payload must be a REAL RFC literal —
# an IPv6 address (ipaddress parse) or an IPvFuture literal
# ("v" + HEXDIG+ + "." + unreserved / sub-delims / ":" only).
# Anything else inside brackets is rejected, closing the payload
# grammar as a class.
if _re.fullmatch(r"v[0-9A-Fa-f]+\.[A-Za-z0-9._~!$&'()*+,;=:-]*", payload):
pass # IPvFuture (case-normalized below)
elif _re.fullmatch(r"[0-9A-Fa-f:.]+", payload):
# strict IPv6 lexical form (hex/colon/dot only — ipaddress alone
# would also accept scoped zone-ids like fe80::1%eth0, which are
# not valid URI host grammar unless %25-encoded)
try:
_ip.IPv6Address(payload)
except ValueError:
err("canonical_remote",
f"bracket payload {payload!r} is not a valid IPv6 address")
else:
err("canonical_remote",
f"bracket payload {payload!r} is neither a valid IPv6 address nor an IPvFuture literal (v+HEXDIG+.+unreserved/sub-delims/colon)")
host = f"[{payload.lower()}]" # brackets preserved (IPv6 + IPvFuture)
else:
# R6-B1: the non-bracketed branch — urlsplit PARSES but does not
# VALIDATE reg-name, and netloc-nonempty is not host presence.
# Split the raw authority ourselves (host[:port]) and validate the
# raw host against real grammar: unreserved / sub-delims / complete
# %HH octets (reg-name), or IPv4 dotted-quad (reg-name's numeric
# case). Port must be all digits. No branch trusts urlsplit alone.
import re as _re
raw_host, sep, raw_port = p.netloc.rpartition(":")
if sep and _re.fullmatch(r"[0-9]+", raw_port):
pass # host:port split
elif sep:
err("canonical_remote", f"invalid port {raw_port!r} in authority {p.netloc!r} (ASCII digits only)")
else:
raw_host, raw_port = p.netloc, None
if not raw_host:
err("canonical_remote", f"empty host in authority {p.netloc!r}")
# strict reg-name / IPv4 scan: unreserved + sub-delims, with '%'
# only inside complete %HH octets (IPv4 dotted-quad is a subset of
# this charset — digits and dots — so one scan covers both).
import re as _re
i = 0
ok_host = True
while i < len(raw_host):
c = raw_host[i]
if c == "%":
if i + 2 >= len(raw_host) or not _re.fullmatch(r"[0-9A-Fa-f]{2}", raw_host[i+1:i+3]):
ok_host = False; break
i += 3
elif c in "!$&'()*+,;=-._~" or ("a" <= c <= "z") or ("A" <= c <= "Z") or ("0" <= c <= "9"):
# R7-B1: EXPLICIT ASCII only — str.isalnum() is Unicode-aware
# and admits non-ASCII letters/digits (é, full-width ). Policy
# is ASCII-only reg-name; punycode xn-- is the sanctioned
# Unicode spelling and remains legal under this charset.
i += 1
else:
ok_host = False; break
if not ok_host:
err("canonical_remote", f"host {raw_host!r} is not valid reg-name/IPv4 grammar (unreserved/sub-delims/complete %HH only)")
host = raw_host.lower()
port = p.port # None when absent; preserved whenever explicitly present (B2: incl. 0)
authority = host + (f":{port}" if port is not None else "")
path = p.path or "/"
# canonical trailing-slash + .git strip as ONE operation (F4): slash
# first, then .git, then any slash exposed by that strip.
path = path.rstrip("/")
if path.endswith(".git"):
path = path[:-4].rstrip("/")
return f"{scheme}://{authority}{path or ''}"
if "--mirror-path" in ARGS:
idx = ARGS.index("--mirror-path")
parts = ARGS[idx + 1:]
if len(parts) != 3:
err("usage", "--mirror-path takes <host> <owner> <repo>")
print(check_mirror_components(*parts))
sys.exit(0)
if "--normalize-remote" in ARGS:
idx = ARGS.index("--normalize-remote")
vals = ARGS[idx + 1:]
if len(vals) != 1:
err("usage", "--normalize-remote takes <url>")
print(normalize_remote(vals[0]))
sys.exit(0)
# ---- arg parsing ----
if not ARGS:
err("usage", "a repo.json path is required")
path = None
i = 0
while i < len(ARGS):
a = ARGS[i]
if a == "--mode":
i += 1
if i >= len(ARGS) or ARGS[i] not in ("managed", "display"):
err("usage", "--mode takes managed|display")
MODE = ARGS[i]
elif a == "--require-v2":
REQUIRE_V2 = True
elif a.startswith("--"):
err("usage", f"unknown option {a}")
else:
if path is not None:
err("usage", "multiple file arguments")
path = a
i += 1
if path is None:
err("usage", "a repo.json path is required")
# ---- load: strict UTF-8, strict RFC JSON (F3) ----
try:
with open(path, "rb") as fh:
raw_bytes = fh.read()
except OSError as e:
err("file", str(e))
try:
raw = raw_bytes.decode("utf-8")
except UnicodeDecodeError as e:
err("json", f"invalid UTF-8: {e}")
def _reject_constant(name):
raise ValueError(f"non-RFC JSON constant {name}")
try:
doc = json.loads(raw, parse_constant=_reject_constant)
except (json.JSONDecodeError, ValueError) as e:
err("json", f"malformed JSON: {e}")
if not isinstance(doc, dict):
err("json", "top level must be an object")
HOST_ROOT = os.environ.get("MOSAIC_HOST_ROOT", "")
V1_KEYS = {"integration_trunk", "release_branch"}
V2_REQUIRED = ["schema_version", "integration_trunk", "release_branch", "flow",
"canonical_remote", "canonical_clone"]
V2_OPTIONAL = {"worktree_root", "worktree_policy", "notes", "x_extensions"}
ENUM_FLOW = {"direct", "trunk-release"}
ENUM_POLICY = {"tool-managed", "orchestrator-precreated"}
def check_branch(key, value):
# Delegate the full git branch grammar to git itself (F2). B1: reject
# reflog shorthand BEFORE delegation — `git check-ref-format --branch
# '@{-n}'` expands from the CALLER repo's checkout history, making
# validation cwd-dependent; a persistent declaration must never bind
# to ambient reflog state.
if not isinstance(value, str) or not value:
err(key, "must be a non-empty string")
if "@{" in value:
err(key, f"{value!r} contains '@{{' reflog/namespace shorthand — declarations must be literal branch names (B1)")
if value.startswith("refs/heads/"): # check-ref-format --branch strips this; we do not allow it
err(key, "bare branch name expected, not a full ref")
try:
r = subprocess.run(["git", "check-ref-format", "--branch", value],
capture_output=True)
except OSError as e:
err(key, f"cannot invoke git check-ref-format: {e}")
if r.returncode != 0:
err(key, f"{value!r} is not a valid git branch name (git check-ref-format, §5.2)")
def check_host_path(key, value):
# §1.2a: host:/-anchored; canonical segment normalization; empty/./.. rejected
# BEFORE resolution; tilde rejected outright.
if not isinstance(value, str) or not value:
err(key, "must be a non-empty string")
if "~" in value:
err(key, "tilde-anchored path rejected (§1.2a: ~ binds to caller HOME)")
if not value.startswith("host:/"):
err(key, "must be host:/-anchored (§1.2a)")
rest = value[len("host:/"):]
if rest == "":
err(key, "no segments after host:/")
segments = rest.split("/")
for seg in segments:
if seg == "":
err(key, f"empty segment in {value!r} (canonical normalization, §1.2a)")
if seg in (".", ".."):
err(key, f"dot segment {seg!r} rejected before resolution (§1.2a)")
return segments
# ---- version ----
sv = doc.get("schema_version")
if "schema_version" in doc:
if not isinstance(sv, int) or isinstance(sv, bool):
err("schema_version", "must be an integer")
if sv not in (1, 2):
err("schema_version", f"unknown schema_version {sv} — treated as ABSENT per keep-list K3; update tooling")
version = sv
else:
version = 1
warn("schema_version absent → v1 compatibility mode (two keys only)")
if REQUIRE_V2 and version != 2:
err("schema_version", "CI authoring rule: new or edited declarations must declare schema_version 2")
# ---- v1 ----
if version == 1:
for k in V1_KEYS:
check_branch(k, doc.get(k))
extra = set(doc) - V1_KEYS
if extra:
err("x_extensions", f"unknown top-level keys in v1: {sorted(extra)}")
print("OK (v1)")
sys.exit(0)
# ---- v2 required ----
for k in V2_REQUIRED:
if k not in doc:
err(k, "required for v2 (§1.2)")
check_branch("integration_trunk", doc["integration_trunk"])
check_branch("release_branch", doc["release_branch"])
# type-check BEFORE membership (F3: list-typed enums must not traceback)
if not isinstance(doc["flow"], str) or doc["flow"] not in ENUM_FLOW:
err("flow", f"must be one of {sorted(ENUM_FLOW)} (§1.2, required — no defaulting, R7)")
unknown = set(doc) - set(V2_REQUIRED) - V2_OPTIONAL
if unknown:
err("x_extensions", f"unknown top-level keys {sorted(unknown)} — place extensions inside x_extensions")
if "worktree_policy" in doc and (not isinstance(doc["worktree_policy"], str)
or doc["worktree_policy"] not in ENUM_POLICY):
err("worktree_policy", f"must be one of {sorted(ENUM_POLICY)}")
if "notes" in doc and not isinstance(doc["notes"], str):
err("notes", "must be a string")
if "x_extensions" in doc and not isinstance(doc["x_extensions"], dict):
err("x_extensions", "must be an object")
for strkey in ("canonical_remote", "canonical_clone", "worktree_root"):
if strkey in doc and not isinstance(doc[strkey], str):
err(strkey, "must be a string")
# ---- remote (§5.3) ----
if not isinstance(doc["canonical_remote"], str):
err("canonical_remote", "must be a string")
else:
normalize_remote(doc["canonical_remote"])
# ---- paths (§1.2a) ----
canonical_segments = check_host_path("canonical_clone", doc["canonical_clone"])
wt_segments = None
if "worktree_root" in doc:
wt_segments = check_host_path("worktree_root", doc["worktree_root"])
# ---- cross-field (§5.2) ----
trunk, rel, flow = doc["integration_trunk"], doc["release_branch"], doc["flow"]
if flow == "direct" and trunk != rel:
err("flow", "direct requires integration_trunk == release_branch (§5.2)")
if flow == "trunk-release" and trunk == rel:
err("flow", "trunk-release requires integration_trunk != release_branch (§5.2)")
# ---- root gate (§1.2a fail-closed; T51P2R1 F1) ----
# Every v2 declaration consumes a path (canonical_clone is required), so
# managed mode cannot proceed without a provable host anchor. Display mode
# warns and omits root-dependent resolution only.
if not HOST_ROOT:
if MODE == "managed":
err("MOSAIC_HOST_ROOT",
"unset or empty — managed validation of a v2 declaration consumes paths "
"(canonical_clone required); fail closed (§1.2a, DR3 X2b)")
else:
warn("host root unset; root-dependent resolution omitted (display mode, §1.2a)")
if doc.get("worktree_policy") == "tool-managed":
if "worktree_root" not in doc:
err("worktree_policy", "tool-managed requires worktree_root (containment provable, §4.5)")
if HOST_ROOT:
root_real = os.path.realpath(HOST_ROOT)
# B3 fix: containment is tested on the LEXICAL normalized path, not
# on realpath of the joined result — a child symlink under the host
# root can no longer fake outside-ness. host:/ segments are always
# lexically under the root, so tool-managed fails universally until
# the anchor scheme grows a real outside-root form (J3 charter).
resolved = os.path.normpath(os.path.join(root_real, *wt_segments))
if resolved == root_real or resolved.startswith(root_real + os.sep):
err("worktree_policy",
f"tool-managed worktree_root resolves inside MOSAIC_HOST_ROOT (§4.5: outside-root requirement)")
# no-root case: managed mode already failed at the gate above; display warned
print("OK")
try:
main()
except SystemExit:
raise
except Exception as e: # F3: no traceback may ever escape the contract
err("internal", f"input rejected (unexpected condition: {type(e).__name__})")
PYEOF
@@ -32,9 +32,7 @@
# 0 delivered (submitted) or queued (agent busy; will process when free)
# 1 tmux target not found
# 2 submission NOT confirmed — either still an unsubmitted draft, or the REPL
# input box could not be located to confirm the message actually landed.
# Locating the box is runtime-specific; see locate_input_box() below, and
# add a shape there before pointing this tool at a new runtime.
# input prompt could not be located to confirm the message actually landed.
# Delivery is NEVER inferred from absence of evidence: if we cannot positively
# see the input box clear of the message (or the queued banner), we fail loud
# so the sender learns immediately instead of a silent worker->lead stall.
@@ -99,50 +97,10 @@ printf '%s' "$MSG" | "${tmux_cmd[@]}" load-buffer -b "$BUF" -
# would otherwise accumulate forever.
sleep 0.5
# Locate the REPL input box in a captured pane. Prints the box's contents on
# stdout and returns 0 when the box was FOUND; returns 1 when it could not be
# located at all. Found-but-empty is a real, distinct answer (an empty input box
# is what a submitted message leaves behind), so the caller must branch on the
# return code, never on whether the output is empty.
#
# Two REPL shapes are recognised:
# * a prompt-glyph line — ``, a leading `>`, or `│ >`. Claude Code and most
# readline REPLs.
# * a box drawn as two horizontal `─` rules with the input between them and NO
# prompt glyph anywhere. pi renders this. Anchoring on the LAST rule pair is
# what makes it safe: agent output can contain its own rules, but nothing is
# drawn below the input box except the status line.
#
# Adding a runtime means adding its shape HERE. A shape that is missing does not
# degrade gracefully: it turns every send to that runtime into a false
# "may be UNDELIVERED", which is what #1362 measured on pi and #1257 on another
# arm of the same probe.
locate_input_box() {
local pane=$1 glyph_line rule_lines top bottom
glyph_line=$(printf '%s\n' "$pane" | grep -E '|^>|│ >' | tail -1)
if [ -n "$glyph_line" ]; then printf '%s\n' "$glyph_line"; return 0; fi
rule_lines=$(printf '%s\n' "$pane" | grep -nE '^[[:space:]]*─{4,}[[:space:]]*$' | cut -d: -f1 | tail -2)
[ -n "$rule_lines" ] || return 1
# Split the (at most two) captured line numbers with parameter expansion. Not
# `head -1`: piping into an early-exiting consumer SIGPIPEs the producer, which
# under `set -euo pipefail` aborts the caller with rc=141 and no output. The
# scripts/pipefail-early-exit.test.mjs guard reds on that shape, correctly.
# With one rule captured both halves resolve to the same value and the
# ordering test below rejects it, which is the answer we want anyway.
top=${rule_lines%%$'\n'*}
bottom=${rule_lines##*$'\n'}
[ "$top" != "$bottom" ] || return 1
[ "$bottom" -gt "$top" ] || return 1
# An empty range (adjacent rules) prints nothing and still returns 0: found,
# empty, which is the delivered shape.
printf '%s\n' "$pane" | sed -n "$((top + 1)),$((bottom - 1))p"
return 0
}
# 2) Submit, then POSITIVELY confirm submission; flush with another Enter if it is
# still a draft. Success requires positive evidence — the queued banner, OR the
# REPL input box located AND clear of our message tail. The historical bug was
# treating ABSENCE of a draft as delivery: if the input box was never located
# treating ABSENCE of a draft as delivery: if the prompt glyph was never matched
# (wrong pane / prompt-glyph drift), an unsubmitted message read as "delivered"
# and worker->lead relays stalled silently. We now default to UNCONFIRMED and only
# upgrade to delivered on positive evidence; anything we cannot confirm fails loud.
@@ -155,14 +113,15 @@ for attempt in $(seq 1 $((RETRIES + 1))); do
if grep -qF "$QUEUED_RE" <<<"$pane"; then
status="queued"; break
fi
# If we cannot see the input box, we have NO evidence of submission state —
# stay UNCONFIRMED and retry; never infer delivery.
if ! inputbox=$(locate_input_box "$pane"); then
# Locate the REPL input box (prompt glyph). If we cannot see it, we have NO
# evidence of submission state — stay UNCONFIRMED and retry; never infer delivery.
promptline=$(printf '%s' "$pane" | grep -E '|^>|│ >' | tail -1)
if [ -z "$promptline" ]; then
status="unconfirmed"; continue
fi
# Input box located AND still carrying our tail => unsubmitted draft. Flush + retry.
# (Submitted messages scroll up into history; a draft stays in the box.)
if [ -n "$snippet" ] && grep -qF "$snippet" <<<"$inputbox"; then
# (Submitted messages scroll up into history; a draft stays on the line.)
if [ -n "$snippet" ] && grep -qF "$snippet" <<<"$promptline"; then
status="draft"; continue
fi
# Input box located AND clear of our tail => positively submitted. This is the
@@ -176,6 +135,6 @@ case "$status" in
delivered) echo "✓ delivered to $TARGET"; exit 0 ;;
queued) echo "✓ queued to $TARGET (agent busy — will process when it returns to prompt)"; exit 0 ;;
draft) echo "✗ still an unsubmitted draft on $TARGET after $RETRIES flush attempts" >&2; exit 2 ;;
unconfirmed) echo "✗ could not confirm submission on $TARGET: REPL input box not locatable after $((RETRIES + 1)) attempts — message may be UNDELIVERED (check target/pane, retry, or escalate)" >&2; exit 2 ;;
unconfirmed) echo "✗ could not confirm submission on $TARGET: REPL input prompt not locatable after $((RETRIES + 1)) attempts — message may be UNDELIVERED (check target/pane, retry, or escalate)" >&2; exit 2 ;;
*) echo "✗ could not confirm submission on $TARGET (unexpected state '$status')" >&2; exit 2 ;;
esac
@@ -10,13 +10,6 @@
# "could not confirm submission").
# 3. DRAFT — a ` `-prompt pane that never submits (message stays on the
# input line) => exit 2, stderr "unsubmitted draft".
# 4. DELIVERED — a pane whose input box is two `─` rules with NO prompt glyph
# (box shape) anywhere (pi's shape) and which submits => exit 0. Pre-#1362
# the glyph probe could not see this box at all, so EVERY send
# to such a pane reported "may be UNDELIVERED" while landing.
# 5. DRAFT — the same glyphless box, holding our tail across every flush
# (box shape) Enter => exit 2, stderr "unsubmitted draft". Pre-#1362 this
# also reported unconfirmed, so the true state was invisible.
set -uo pipefail
HERE=$(cd -- "$(dirname -- "$0")" && pwd)
@@ -76,56 +69,6 @@ else
fi
fi
# --- Fixtures 4 and 5: a pi-shaped pane. The input box is two `─` rules with the
# text between them and NO prompt glyph anywhere, so the glyph probe alone can
# never locate it and every send reports "may be UNDELIVERED" (#1362). The
# renderer below is the shape, not the runtime: MODE=clear submits (box empties),
# MODE=keep leaves the text sitting in the box.
cat > "$TMP/pibox.sh" <<'PIBOX'
#!/usr/bin/env bash
MODE=${1:-clear}
RULE=$(printf '─%.0s' $(seq 1 60))
buf=""
draw() {
printf '\033[H\033[2J'
printf 'fixture output line\n\n'
printf '%s\n' "$RULE"
printf '%s\n' "$buf"
printf '%s\n' "$RULE"
printf '~/fixture (main)\n'
printf 'tok 0 model fixture\n'
}
draw
while IFS= read -r line; do
# keep: hold the tail across every flush Enter, which is what a stuck draft does.
if [ "$MODE" = keep ]; then [ -n "$line" ] && buf=$line; else buf=""; fi
draw
done
PIBOX
chmod +x "$TMP/pibox.sh"
tmux -L "$SOCKET" new-session -d -s pibox -c "$TMP" "exec bash '$TMP/pibox.sh' clear"
sleep 0.3
out=$("$SEND" -L "$SOCKET" -t "=pibox" -m "pi fixture four delivered ok" 2>"$TMP/e4"); rc=$?
if [ "$rc" -eq 0 ] && grep -qF "✓ delivered" <<<"$out"; then
ok "delivered: glyphless box-drawn REPL that submits => exit 0 ✓ delivered"
else
no "delivered: glyphless box-drawn REPL that submits => exit 0 ✓ delivered" "rc=$rc out=[$out] err=[$(cat "$TMP/e4")]"
fi
tmux -L "$SOCKET" new-session -d -s piboxdraft -c "$TMP" "exec bash '$TMP/pibox.sh' keep"
sleep 0.3
if out=$("$SEND" -L "$SOCKET" -t "=piboxdraft" -r 1 -m "pi fixture five stuck in the box" 2>"$TMP/e5"); then
no "draft: glyphless box-drawn pane holding our tail must NOT report success" "expected exit 2, got 0 (out=[$out])"
else
rc=$?
if [ "$rc" -eq 2 ] && grep -qF "unsubmitted draft" "$TMP/e5"; then
ok "draft: message left in a glyphless box => exit 2 + 'unsubmitted draft'"
else
no "draft: message left in a glyphless box => exit 2 + stderr" "rc=$rc err=[$(cat "$TMP/e5")]"
fi
fi
echo "---"
echo "PASS=$PASS FAIL=$FAIL"
[ "$FAIL" -eq 0 ]
+1 -1
View File
@@ -25,7 +25,7 @@
"lint": "eslint src",
"typecheck": "tsc --noEmit",
"test": "vitest run --passWithNoTests && pnpm run test:framework-shell",
"test:framework-shell": "bash framework/tools/quality/scripts/check-test-enumeration.sh && bash framework/tools/quality/scripts/test-check-test-enumeration.sh && python3 framework/tools/quality/scripts/test-framework-drift-check.py && bash framework/tools/quality/scripts/test-framework-drift-doctor.sh && bash framework/systemd/user/test-fleet-units.sh && python3 src/lease-broker/daemon_deadline_unittest.py && python3 src/lease-broker/normative_fragments_unittest.py && python3 src/lease-broker/promotion_binding_unittest.py && python3 src/lease-broker/promotion_trigger_unittest.py && python3 src/lease-broker/receipt_challenge_unittest.py && python3 src/lease-broker/context_recovery_unittest.py && python3 src/lease-broker/recovery_runtime_unittest.py && python3 src/lease-broker/recovery_b1_adversarial_unittest.py && python3 src/lease-broker/receipt_observer_client_unittest.py && python3 src/lease-broker/invariant_r_unittest.py && python3 src/lease-broker/framework_skill_portability_unittest.py && python3 src/lease-broker/revoke_noop_unittest.py && python3 src/mutator-gate/runtime_tools_unittest.py && python3 src/mutator-gate/runtime_launch_guard_unittest.py && python3 src/mutator-gate/version_coupling_unittest.py && python3 framework/tools/lease-broker/check-runtime-launches.py --root ../.. && bash framework/tools/codex/test-pr-diff-context.sh && bash framework/tools/qa/test-deps-preflight.sh && bash framework/tools/git/test-pr-edit.sh && bash framework/tools/git/test-pr-create-fallback-default-base.sh && bash framework/tools/git/test-pr-review-gitea-comment.sh && bash framework/tools/git/test-pr-review-repo-host-override.sh && bash framework/tools/git/test-ci-queue-wait-no-status.sh && bash framework/tools/git/test-ci-queue-wait-branch-absent.sh && bash framework/tools/git/test-ci-queue-wait-tristate.sh && bash framework/tools/git/test-ci-queue-wait-github-checks.sh && bash framework/tools/git/test-ci-queue-wait-no-ci-expected.sh && bash framework/tools/git/test-pr-merge-queue-branch.sh && bash framework/tools/git/test-pr-merge-no-ci-expected.sh && bash framework/tools/git/test-pr-merge-fork-ci-status.sh && bash framework/tools/git/test-pr-merge-head-pin.sh && bash framework/tools/git/test-pr-merge-message-field.sh && bash framework/tools/git/test-git-credential-mosaic.sh && bash framework/tools/git/test-gitea-token-identity.sh && bash framework/tools/git/test-explain-diagnostic-status-neutral.sh && bash framework/tools/git/test-detect-platform-outside-repo.sh && bash framework/tools/woodpecker/test-terminal-green-contract.sh && bash framework/tools/_scripts/test-install-ordering-guard.sh && bash framework/tools/_scripts/test-mosaic-init-rce.sh && bash framework/tools/tmux/agent-send.test.sh && bash framework/tools/wake/test-wake-store-ack.sh && bash framework/tools/wake/test-wake-store-enqueue-race.sh && bash framework/tools/wake/test-wake-digest-hmac.sh && bash framework/tools/wake/test-wake-digest-quarantine.sh && bash framework/tools/wake/test-wake-detector.sh && bash framework/tools/wake/test-wake-fn-oracle.sh && bash framework/tools/wake/test-wake-reconcile.sh && bash framework/tools/wake/test-wake-beacon.sh && bash framework/tools/wake/test-wake-preimage.sh && bash framework/tools/wake/test-wake-install.sh && bash framework/tools/glpi/test-list-http-status.sh && bash framework/tools/orchestrator/test-board-roll.sh && bash framework/tools/woodpecker/test-ci-wait-exit-matrix.sh && bash framework/tools/_scripts/test-fleet-transport-check.sh && bash framework/tools/_scripts/test-brain-home-check.sh && bash framework/tools/fleet/test-agent-session-broker-preflight.sh"
"test:framework-shell": "bash framework/tools/quality/scripts/check-test-enumeration.sh && bash framework/tools/quality/scripts/test-check-test-enumeration.sh && python3 framework/tools/quality/scripts/test-framework-drift-check.py && bash framework/tools/quality/scripts/test-framework-drift-doctor.sh && bash framework/systemd/user/test-fleet-units.sh && python3 src/lease-broker/daemon_deadline_unittest.py && python3 src/lease-broker/normative_fragments_unittest.py && python3 src/lease-broker/promotion_binding_unittest.py && python3 src/lease-broker/promotion_trigger_unittest.py && python3 src/lease-broker/receipt_challenge_unittest.py && python3 src/lease-broker/context_recovery_unittest.py && python3 src/lease-broker/recovery_runtime_unittest.py && python3 src/lease-broker/recovery_b1_adversarial_unittest.py && python3 src/lease-broker/receipt_observer_client_unittest.py && python3 src/lease-broker/invariant_r_unittest.py && python3 src/lease-broker/framework_skill_portability_unittest.py && python3 src/lease-broker/revoke_noop_unittest.py && python3 src/mutator-gate/runtime_tools_unittest.py && python3 src/mutator-gate/runtime_launch_guard_unittest.py && python3 src/mutator-gate/version_coupling_unittest.py && python3 framework/tools/lease-broker/check-runtime-launches.py --root ../.. && bash framework/tools/codex/test-pr-diff-context.sh && bash framework/tools/qa/test-deps-preflight.sh && bash framework/tools/git/test-pr-edit.sh && bash framework/tools/git/test-pr-review-gitea-comment.sh && bash framework/tools/git/test-pr-review-repo-host-override.sh && bash framework/tools/git/test-ci-queue-wait-no-status.sh && bash framework/tools/git/test-ci-queue-wait-branch-absent.sh && bash framework/tools/git/test-ci-queue-wait-tristate.sh && bash framework/tools/git/test-ci-queue-wait-github-checks.sh && bash framework/tools/git/test-pr-merge-queue-branch.sh && bash framework/tools/git/test-pr-merge-fork-ci-status.sh && bash framework/tools/git/test-pr-merge-head-pin.sh && bash framework/tools/git/test-pr-merge-message-field.sh && bash framework/tools/git/test-git-credential-mosaic.sh && bash framework/tools/git/test-gitea-token-identity.sh && bash framework/tools/git/test-explain-diagnostic-status-neutral.sh && bash framework/tools/git/test-detect-platform-outside-repo.sh && bash framework/tools/woodpecker/test-terminal-green-contract.sh && bash framework/tools/_scripts/test-install-ordering-guard.sh && bash framework/tools/_scripts/test-mosaic-init-rce.sh && bash framework/tools/tmux/agent-send.test.sh && bash framework/tools/wake/test-wake-store-ack.sh && bash framework/tools/wake/test-wake-store-enqueue-race.sh && bash framework/tools/wake/test-wake-digest-hmac.sh && bash framework/tools/wake/test-wake-digest-quarantine.sh && bash framework/tools/wake/test-wake-detector.sh && bash framework/tools/wake/test-wake-fn-oracle.sh && bash framework/tools/wake/test-wake-reconcile.sh && bash framework/tools/wake/test-wake-beacon.sh && bash framework/tools/wake/test-wake-preimage.sh && bash framework/tools/wake/test-wake-install.sh && bash framework/tools/glpi/test-list-http-status.sh && bash framework/tools/orchestrator/test-board-roll.sh && bash framework/tools/woodpecker/test-ci-wait-exit-matrix.sh && bash framework/tools/_scripts/test-fleet-transport-check.sh && bash framework/tools/_scripts/test-brain-home-check.sh"
},
"dependencies": {
"@mosaicstack/brain": "workspace:*",
@@ -1,177 +0,0 @@
import { lstat, mkdir, mkdtemp, readFile, rm, symlink, writeFile } from 'node:fs/promises';
import { tmpdir } from 'node:os';
import { join } from 'node:path';
import { afterEach, describe, expect, it } from 'vitest';
import { placeUnitFile, resolveLeaseBrokerSocketForPreflight } from './fleet.js';
/**
* Unit-placement regression harness for #1292.
*
* The two measured defects this suite pins:
* 1. `systemctl enable <name>` does NOT rewrite an existing by-path
* wants-symlink — so placement must remove stale residue explicitly, and
* acceptance asserts on the RESULTING SYMLINK TARGET, never on the enable
* call's argument (asserting the call cannot see where the link ended up).
* 2. Node's copyFile FOLLOWS a by-path symlink at the destination and
* overwrites the SEED template. Acceptance asserts on the SEED's bytes
* AND mtime — unchanged — which is the only check that can redden for
* finding 2. The symlink-target assertion catches finding 1; these are
* different defects with different failure modes.
*
* Fixtures are entirely inside tmpdirs (source template, active systemd dir,
* wants dir) — no real host paths are touched by this suite.
*/
describe('placeUnitFile (#1292 unit placement)', () => {
const cleanup: string[] = [];
afterEach(async () => {
while (cleanup.length > 0) {
await rm(cleanup.pop()!, { recursive: true, force: true });
}
});
async function fixture() {
const root = await mkdtemp(join(tmpdir(), 'place-unit-'));
cleanup.push(root);
const seedDir = join(root, 'seed');
const activeDir = join(root, 'active');
await mkdir(seedDir, { recursive: true });
await mkdir(activeDir, { recursive: true });
const seedTemplate = join(seedDir, 'unit-under-test.service');
await writeFile(
seedTemplate,
'[Unit]\nDescription=seed template\n[Service]\nType=oneshot\nExecStart=/bin/true\n[Install]\nWantedBy=default.target\n',
);
const activeSource = join(root, 'active-source.service');
await writeFile(
activeSource,
'[Unit]\nDescription=active copy v2\n[Service]\nType=oneshot\nExecStart=/bin/true\n[Install]\nWantedBy=default.target\n',
);
return { root, seedDir, activeDir, seedTemplate, activeSource };
}
it('places a regular file on a clean host (negative control: no residue anywhere)', async () => {
const f = await fixture();
const result = await placeUnitFile(f.activeSource, f.activeDir, 'unit-under-test.service');
expect(result.unlinkedDestinationSymlink).toBe(false);
expect(result.removedStaleWantsSymlink).toBe(false);
const info = await lstat(join(f.activeDir, 'unit-under-test.service'));
expect(info.isSymbolicLink()).toBe(false);
expect(await readFile(join(f.activeDir, 'unit-under-test.service'), 'utf8')).toContain(
'active copy v2',
);
// Seed untouched by construction — but assert it, so the clean-host case
// cannot silently regress into seed-mutation.
expect(await readFile(f.seedTemplate, 'utf8')).toContain('seed template');
});
it('by-path residue: unlinks destination symlink, places the file, seed bytes AND mtime unchanged (finding 2)', async () => {
const f = await fixture();
const seedBefore = await readFile(f.seedTemplate, 'utf8');
const mtimeBefore = (await lstat(f.seedTemplate)).mtimeMs;
// The fomo-lin convention: by-path enable left a symlink AT the unit name
// pointing at the seed template, plus a wants-symlink doing the same.
await symlink(f.seedTemplate, join(f.activeDir, 'unit-under-test.service'));
const wantsDir = join(f.activeDir, 'default.target.wants');
await mkdir(wantsDir, { recursive: true });
await symlink(f.seedTemplate, join(wantsDir, 'unit-under-test.service'));
const result = await placeUnitFile(f.activeSource, f.activeDir, 'unit-under-test.service');
expect(result.unlinkedDestinationSymlink).toBe(true);
expect(result.removedStaleWantsSymlink).toBe(true);
// FINDING 2's check: the seed is byte-identical and its mtime did not move.
expect(await readFile(f.seedTemplate, 'utf8')).toBe(seedBefore);
expect((await lstat(f.seedTemplate)).mtimeMs).toBe(mtimeBefore);
// The destination is now a regular file carrying the ACTIVE content.
const destInfo = await lstat(join(f.activeDir, 'unit-under-test.service'));
expect(destInfo.isSymbolicLink()).toBe(false);
expect(await readFile(join(f.activeDir, 'unit-under-test.service'), 'utf8')).toContain(
'active copy v2',
);
});
it('by-path residue: no wants-symlink remains pointing at the seed (finding 1 residue cleared)', async () => {
const f = await fixture();
await symlink(f.seedTemplate, join(f.activeDir, 'unit-under-test.service'));
const wantsDir = join(f.activeDir, 'default.target.wants');
await mkdir(wantsDir, { recursive: true });
await symlink(f.seedTemplate, join(wantsDir, 'unit-under-test.service'));
await placeUnitFile(f.activeSource, f.activeDir, 'unit-under-test.service');
// After placement the stale wants link is GONE (enable-by-name recreates
// it correctly). A link still present must not point at the seed.
try {
const link = await lstat(join(wantsDir, 'unit-under-test.service'));
if (link.isSymbolicLink()) {
const target = await readFile(join(wantsDir, 'unit-under-test.service'), 'utf8').catch(
async () => '',
);
expect(target).not.toContain('seed template');
}
} catch {
// absent wants link — the expected post-placement state
}
});
it('idempotence: second placement on a reconciled host is a no-op producing the identical final state', async () => {
const f = await fixture();
// Reconciled starting state: regular file at the name, wants link to the active copy.
await writeFile(
join(f.activeDir, 'unit-under-test.service'),
await readFile(f.activeSource, 'utf8'),
);
const wantsDir = join(f.activeDir, 'default.target.wants');
await mkdir(wantsDir, { recursive: true });
await symlink(
join(f.activeDir, 'unit-under-test.service'),
join(wantsDir, 'unit-under-test.service'),
);
const before = await readFile(join(f.activeDir, 'unit-under-test.service'), 'utf8');
const result = await placeUnitFile(f.activeSource, f.activeDir, 'unit-under-test.service');
// No destructive step fired: no unlink, no wants removal.
expect(result.unlinkedDestinationSymlink).toBe(false);
expect(result.removedStaleWantsSymlink).toBe(false);
// Identical final state.
expect(await readFile(join(f.activeDir, 'unit-under-test.service'), 'utf8')).toBe(before);
const link = await lstat(join(wantsDir, 'unit-under-test.service'));
expect(link.isSymbolicLink()).toBe(true);
});
it('double install on by-path residue converges to the identical reconciled state', async () => {
const f = await fixture();
await symlink(f.seedTemplate, join(f.activeDir, 'unit-under-test.service'));
const wantsDir = join(f.activeDir, 'default.target.wants');
await mkdir(wantsDir, { recursive: true });
await symlink(f.seedTemplate, join(wantsDir, 'unit-under-test.service'));
await placeUnitFile(f.activeSource, f.activeDir, 'unit-under-test.service');
const first = await readFile(join(f.activeDir, 'unit-under-test.service'), 'utf8');
const secondRun = await placeUnitFile(f.activeSource, f.activeDir, 'unit-under-test.service');
const second = await readFile(join(f.activeDir, 'unit-under-test.service'), 'utf8');
expect(secondRun.unlinkedDestinationSymlink).toBe(false);
expect(second).toBe(first);
});
});
describe('resolveLeaseBrokerSocketForPreflight (#1292 preflight resolution)', () => {
it('explicit MOSAIC_LEASE_BROKER_SOCKET wins', () => {
expect(
resolveLeaseBrokerSocketForPreflight({ MOSAIC_LEASE_BROKER_SOCKET: '/custom/sock' }, 1000),
).toBe('/custom/sock');
});
it('XDG_RUNTIME_DIR next', () => {
expect(resolveLeaseBrokerSocketForPreflight({ XDG_RUNTIME_DIR: '/run/user/1001' }, 1000)).toBe(
'/run/user/1001/mosaic-lease/broker.sock',
);
});
it('falls back to /run/user/<uid>', () => {
expect(resolveLeaseBrokerSocketForPreflight({}, 1002)).toBe(
'/run/user/1002/mosaic-lease/broker.sock',
);
});
});
@@ -73,10 +73,6 @@ function program(
runner,
reconcileDeps: {
homeDirectory: '/home/mosaic',
// Deterministic broker presence: without a seam the reconciler probes the
// REAL host socket (#1297 F3), making every CLI start test answer the
// host's broker state instead of its own property.
checkBrokerSocket: async () => true,
readHolderIdentity: async () => '11111111-1111-4111-8111-111111111111',
validateRoster: async () => undefined,
prepareProjections: async () => [{ agentName: 'coder0' }],
+2 -193
View File
@@ -1,24 +1,11 @@
import {
chmod,
lstat,
mkdir,
mkdtemp,
readFile,
readlink,
rm,
stat,
symlink,
writeFile,
} from 'node:fs/promises';
import { chmod, lstat, mkdir, mkdtemp, readFile, rm, stat, writeFile } from 'node:fs/promises';
import { tmpdir } from 'node:os';
import { dirname, join, resolve } from 'node:path';
import { createServer } from 'node:net';
import { Command } from 'commander';
import { afterEach, describe, expect, it, vi } from 'vitest';
import {
acquireRestartLock,
addAgentToRoster,
brokerSocketPresent,
buildAgentSendCommand,
buildAgentWatchAttachCommand,
buildAgentWatchCommand,
@@ -55,7 +42,6 @@ import {
parseSystemdShow,
parseTmuxListPanes,
parseTmuxListSessions,
placeUnitFile,
registerFleetCommand,
removeAgentFromRoster,
resolveFleetPaths,
@@ -64,7 +50,6 @@ import {
RESTART_LOCK_STALE_MS,
RUNTIME_ACCEPTABLE_COMMANDS,
serializeRosterToYaml,
UnitPlacementError,
VERIFY_DEFAULT_TIMEOUT_MS,
VERIFY_POLL_INTERVAL_MS,
type AgentPsRow,
@@ -851,25 +836,13 @@ describe('fleet command construction', () => {
};
const program = new Command();
program.exitOverride();
// #1292: inject a present broker socket so the preflight passes and this
// spec keeps testing its ORIGINAL property (holder-before-agent ordering).
// The preflight's own refusal behavior has dedicated specs below.
registerFleetCommand(program, {
runner,
mosaicHome: home,
checkBrokerSocket: async () => true,
});
registerFleetCommand(program, { runner, mosaicHome: home });
try {
await program.parseAsync(['node', 'mosaic', 'fleet', 'start']);
await program.parseAsync(['node', 'mosaic', 'fleet', 'stop']);
expect(calls).toEqual([
// #1292: fleet start enables + starts the broker FIRST (enable is
// idempotent; the unit exists after install), re-checking the socket
// before any holder/agent lifecycle effect.
['systemctl', '--user', 'enable', 'mosaic-lease-broker.service'],
['systemctl', '--user', 'start', 'mosaic-lease-broker.service'],
['systemctl', '--user', 'start', 'mosaic-tmux-holder.service'],
['systemctl', '--user', 'start', '[email protected]'],
['systemctl', '--user', 'stop', '[email protected]'],
@@ -880,92 +853,6 @@ describe('fleet command construction', () => {
}
});
it('fleet start refuses with a named error when the broker socket does not appear (#1292)', async () => {
const home = await tempDir();
const rosterPath = join(home, 'fleet', 'roster.yaml');
await mkdir(join(home, 'fleet'), { recursive: true });
await writeFile(
rosterPath,
['version: 1', 'transport: tmux', 'agents:', ' - name: coder0', ' runtime: codex'].join(
'\n',
),
);
const calls: string[][] = [];
const runner: CommandRunner = async (command, args) => {
calls.push([command, ...args]);
return { stdout: '', stderr: '', exitCode: 0 };
};
const program = new Command();
program.exitOverride();
const errors: string[] = [];
const origError = console.error;
console.error = (...args: unknown[]) => {
errors.push(args.join(' '));
};
registerFleetCommand(program, {
runner,
mosaicHome: home,
checkBrokerSocket: async () => false,
});
try {
await program.parseAsync(['node', 'mosaic', 'fleet', 'start']);
// Refused: no holder/agent starts were issued after the broker attempt.
expect(calls).toEqual([
['systemctl', '--user', 'enable', 'mosaic-lease-broker.service'],
['systemctl', '--user', 'start', 'mosaic-lease-broker.service'],
]);
expect(errors.join('\n')).toContain('broker-absent');
expect(errors.join('\n')).toContain('mosaic fleet install');
} finally {
console.error = origError;
await rm(home, { recursive: true, force: true });
}
});
it('fleet start re-probes the broker on the SECOND invocation — no ActiveState trust (#1292 sticky half)', async () => {
const home = await tempDir();
const rosterPath = join(home, 'fleet', 'roster.yaml');
await mkdir(join(home, 'fleet'), { recursive: true });
await writeFile(
rosterPath,
['version: 1', 'transport: tmux', 'agents:', ' - name: coder0', ' runtime: codex'].join(
'\n',
),
);
const calls: string[][] = [];
const runner: CommandRunner = async (command, args) => {
calls.push([command, ...args]);
return { stdout: '', stderr: '', exitCode: 0 };
};
const program = new Command();
program.exitOverride();
// Broker socket NEVER appears — the second start must refuse exactly like
// the first; RemainAfterExit-style stale unit state changes nothing
// because the check is the socket, not systemctl.
registerFleetCommand(program, {
runner,
mosaicHome: home,
checkBrokerSocket: async () => false,
});
const errors: string[] = [];
const origError = console.error;
console.error = (...args: unknown[]) => {
errors.push(args.join(' '));
};
try {
await program.parseAsync(['node', 'mosaic', 'fleet', 'start']);
await program.parseAsync(['node', 'mosaic', 'fleet', 'start']);
// Two invocations, each refusing after its own broker attempt:
expect(
calls.filter((c) => c.join(' ') === 'systemctl --user start [email protected]'),
).toHaveLength(0);
expect(errors.filter((e) => e.includes('broker-absent')).length).toBeGreaterThanOrEqual(2);
} finally {
console.error = origError;
await rm(home, { recursive: true, force: true });
}
});
it('waits for an in-flight restart to clear before relaunching (re-entry guard)', async () => {
const home = await tempDir();
const rosterPath = join(home, 'fleet', 'roster.yaml');
@@ -2179,19 +2066,8 @@ describe('fleet install — auto-enable units for boot-survival', () => {
await enableFleetUnits(runner, minimalRoster, {});
expect(calls).toContainEqual(['systemctl', '--user', 'enable', 'mosaic-lease-broker.service']);
expect(calls).toContainEqual(['systemctl', '--user', 'enable', 'mosaic-tmux-holder.service']);
expect(calls).toContainEqual(['systemctl', '--user', 'enable', '[email protected]']);
// The broker must be enabled BEFORE the holder and agents: a start of any
// gated runtime without the broker is exactly the #1292 4-second death.
const brokerIndex = calls.findIndex(
(c) => c.join(' ') === 'systemctl --user enable mosaic-lease-broker.service',
);
const holderIndex = calls.findIndex(
(c) => c.join(' ') === 'systemctl --user enable mosaic-tmux-holder.service',
);
expect(brokerIndex).toBeGreaterThanOrEqual(0);
expect(brokerIndex).toBeLessThan(holderIndex);
});
it('install still succeeds when systemctl enable returns non-zero (non-fatal)', async () => {
@@ -4486,70 +4362,3 @@ describe('fleet ps — heartbeat path resolution', () => {
);
});
});
describe('#1297 review: the real broker probe, exercised without any seam', () => {
it('brokerSocketPresent answers a REAL unix socket via stat().isSocket() (access(S_IFSOCK) threw ERR_OUT_OF_RANGE)', async () => {
const dir = await tempDir();
const sockPath = join(dir, 'broker.sock');
const server = createServer();
await new Promise<void>((resolve) => {
server.listen(sockPath, resolve);
});
try {
// A live unix socket answers true through the REAL probe — no seam.
expect(await brokerSocketPresent({}, { MOSAIC_LEASE_BROKER_SOCKET: sockPath })).toBe(true);
// Discrimination is by file type: a regular file that EXISTS is not a
// socket. The old implementation could not reach either verdict — it
// threw ERR_OUT_OF_RANGE (node >= 24) and the catch answered false.
const notASocket = join(dir, 'not-a-sock');
await writeFile(notASocket, 'x');
expect(await brokerSocketPresent({}, { MOSAIC_LEASE_BROKER_SOCKET: notASocket })).toBe(false);
// Absent path: false, not a throw.
expect(
await brokerSocketPresent({}, { MOSAIC_LEASE_BROKER_SOCKET: join(dir, 'gone.sock') }),
).toBe(false);
} finally {
await new Promise<void>((resolve) => {
server.close(() => resolve());
});
}
await rm(dir, { recursive: true, force: true });
});
// EACCES-based unlink failure requires a non-root uid: root bypasses
// directory mode bits (CAP_DAC_OVERRIDE), so the abort path cannot be
// triggered this way under CI's root runner. Skipped there, exercised on
// every non-root dev host.
const itUnlessRoot =
typeof process.getuid === 'function' && process.getuid() === 0 ? it.skip : it;
itUnlessRoot(
'placeUnitFile aborts with UnitPlacementError when unlink fails — never copies through a live symlink',
async () => {
const dir = await tempDir();
const unitDir = join(dir, 'systemd', 'user');
await mkdir(unitDir, { recursive: true });
// By-path residue: destination is a symlink pointing somewhere else.
const residueTarget = join(dir, 'residue-target');
await writeFile(residueTarget, 'RESIDUE-BYTES');
const destination = join(unitDir, 'x.service');
await symlink(residueTarget, destination);
const source = join(dir, 'seed.service');
await writeFile(source, 'UNIT-BYTES');
// Read-only unit dir: unlink now fails EACCES (test runs as the owner,
// not root, so mode bits are enforced).
await chmod(unitDir, 0o500);
try {
await expect(placeUnitFile(source, unitDir, 'x.service')).rejects.toThrow(
UnitPlacementError,
);
} finally {
await chmod(unitDir, 0o700);
}
// The copy-through never happened: residue bytes intact, destination
// still the symlink (abort, not overwrite-through).
expect(await readFile(residueTarget, 'utf8')).toBe('RESIDUE-BYTES');
expect(await readlink(destination)).toBe(residueTarget);
await rm(dir, { recursive: true, force: true });
},
);
});
+14 -259
View File
@@ -1,13 +1,11 @@
import { constants, type Stats } from 'node:fs';
import { constants } from 'node:fs';
import {
access,
chmod,
copyFile,
lstat,
mkdir,
open,
readFile,
readlink,
stat,
unlink,
writeFile,
@@ -92,8 +90,6 @@ export type SleepFn = (ms: number) => Promise<void>;
export interface FleetCommandDeps {
runner?: CommandRunner;
/** Test seam for the #1292 fleet-start broker preflight (socket presence). */
checkBrokerSocket?: (path: string) => Promise<boolean> | boolean;
/** Injectable interactive runner for commands needing inherited TTY (e.g., `tmux attach`). */
interactiveRunner?: InteractiveRunner;
/**
@@ -819,135 +815,6 @@ export function buildSystemdEnableCommand(unit: string): string[] {
return ['systemctl', '--user', 'enable', unit];
}
/**
* Place a unit file into the ACTIVE systemd user directory, never through a
* symlink (#1292, measured 2026-08-17).
*
* ⚠ SET-INDEPENDENCE (fomo-lin, 2026-08-17): the set of unit names carrying
* by-path residue and the set of unit names this install copies are
* INDEPENDENT. Until 0.0.50 they were disjoint only by accident of which
* units the install happened to name — fomo-lin survived copy-through solely
* because its one by-path symlink (the broker) was the one unit the install
* did NOT copy. Adding the broker to the copy set made the intersection
* non-empty on the first run. Whoever adds a fifth unit to the placement
* list inherits this helper and its unlink step; do not place units with a
* bare copyFile.
*
* A host provisioned by the enable-by-path convention carries a symlink AT
* the unit-name path in ~/.config/systemd/user/ pointing at the shipped
* template under ~/.config/mosaic/systemd/user/. Node's copyFile FOLLOWS
* that link and overwrites the SEED template instead of placing the active
* unit (verified with fs.copyFile on a throwaway systemd user instance) —
* silent, rc=0, and it mutates the directory every later reseed reads from.
* The same measurement showed `systemctl enable <name>` does NOT rewrite an
* existing by-path wants-symlink, so reconciliation must be explicit.
*
* Placement therefore: if the destination is a symlink, unlink it first
* (unlink → copy — copy-then-unlink would mutate the seed and then destroy
* the evidence that it did); then copy. Also removes a stale
* `default.target.wants/<name>` symlink that points outside the active
* directory (readlink — NOT readFile, which follows the link and returns the
* target's CONTENT), so the subsequent enable-by-name recreates it against
* the active copy. Idempotent: on a clean or already-reconciled destination
* every step is a no-op (the copy rewrites identical bytes).
*
* Returns what was done, for assertions and install reporting.
*/
export interface PlaceUnitResult {
readonly unit: string;
readonly destination: string;
/** A symlink at the unit-name path was unlinked (by-path residue). */
readonly unlinkedDestinationSymlink: boolean;
/** A stale wants-symlink pointing outside the active dir was removed. */
readonly removedStaleWantsSymlink: boolean;
}
/**
* placeUnitFile failed. Thrown BEFORE any copy: no destination bytes were
* written, so a residue target cannot have been clobbered by a copy-through
* (#1297 review F2).
*/
export class UnitPlacementError extends Error {
constructor(
readonly unit: string,
message: string,
) {
super(message);
this.name = UnitPlacementError.name;
}
}
function isErrnoException(error: unknown): error is NodeJS.ErrnoException {
return error instanceof Error && 'code' in error && typeof error.code === 'string';
}
export async function placeUnitFile(
source: string,
systemdUserDir: string,
unit: string,
): Promise<PlaceUnitResult> {
const destination = join(systemdUserDir, unit);
let unlinkedDestinationSymlink = false;
// Destination-absent and unlink-FAILED are different outcomes and must not
// share a catch (#1297 review F2): a swallowed unlink error used to fall
// through to copyFile through the still-live symlink, silently reintroducing
// the exact copy-through this helper exists to prevent.
let destinationInfo: Stats | undefined;
try {
destinationInfo = await lstat(destination);
} catch (error) {
if (!isErrnoException(error) || error.code !== 'ENOENT') {
throw new UnitPlacementError(
unit,
`cannot inspect destination ${destination}: ${error instanceof Error ? error.message : String(error)}`,
);
}
// ENOENT: absent destination — nothing to unlink, copy below is safe.
}
if (destinationInfo?.isSymbolicLink()) {
try {
await unlink(destination);
} catch (error) {
// Abort BEFORE the copy: proceeding would run copyFile through the
// still-live symlink and overwrite the residue target's bytes.
throw new UnitPlacementError(
unit,
`cannot unlink destination symlink ${destination}: ${error instanceof Error ? error.message : String(error)}`,
);
}
unlinkedDestinationSymlink = true;
}
await copyFile(source, destination);
let removedStaleWantsSymlink = false;
const wantsLink = join(systemdUserDir, 'default.target.wants', unit);
try {
const wantsInfo = await lstat(wantsLink);
if (wantsInfo.isSymbolicLink()) {
// readlink — NOT readFile: readFile FOLLOWS the link and returns the
// target file's CONTENT, which is not the question being asked.
let target: string | undefined;
try {
target = await readlink(wantsLink);
} catch {
target = undefined;
}
// Normalize (systemctl writes absolute targets; a relative one resolves
// against the wants dir). A wants-symlink pointing anywhere other than
// the active copy (the by-path convention points at the seed template)
// survives enable-by-name unchanged — remove it so enable recreates it.
if (target !== undefined && resolve(dirname(wantsLink), target) !== destination) {
await unlink(wantsLink);
removedStaleWantsSymlink = true;
}
}
} catch {
// absent wants link — nothing to reconcile
}
return { unit, destination, unlinkedDestinationSymlink, removedStaleWantsSymlink };
}
/**
* Returns the systemctl --user disable command for a given unit.
* Used by `fleet remove` so a removed agent's enabled unit cannot resurrect on
@@ -982,22 +849,6 @@ export async function enableFleetUnits(
let succeeded = 0;
let failed = 0;
// The lease broker ships with the fleet and every gated runtime needs it
// (#1292): seats die at lease registration without it, and no documented
// path ever enabled it. Enabled first — alongside the holder — and the
// unit must have been placed by installFleet's placeUnitFile step.
const brokerResult = await runner(
...splitCommand(buildSystemdEnableCommand('mosaic-lease-broker.service')),
);
if (brokerResult.exitCode === 0) {
succeeded++;
} else {
failed++;
process.stderr.write(
`Warning: could not enable mosaic-lease-broker.service: ${brokerResult.stderr || brokerResult.stdout || 'non-zero exit'}\n`,
);
}
const holderResult = await runner(
...splitCommand(buildSystemdEnableCommand('mosaic-tmux-holder.service')),
);
@@ -1694,7 +1545,7 @@ export function registerFleetCommand(program: Command, deps: FleetCommandDeps =
.description('Install local fleet tools and user systemd units')
.option('--no-enable', 'Skip enabling units for boot-survival')
.action(async (opts: { enable?: boolean }) => {
await installFleet(cmd, frameworkRoot, runner);
await installFleet(cmd, frameworkRoot);
// Unit enablement needs agent names only, so it reads either version.
const roster = await loadRosterReadModel(cmd);
await enableFleetUnits(runner, roster, opts);
@@ -1705,7 +1556,7 @@ export function registerFleetCommand(program: Command, deps: FleetCommandDeps =
.description('Install local fleet tools and user systemd units')
.option('--no-enable', 'Skip enabling units for boot-survival')
.action(async (opts: { enable?: boolean }) => {
await installFleet(cmd, frameworkRoot, runner);
await installFleet(cmd, frameworkRoot);
// Unit enablement needs agent names only, so it reads either version.
const roster = await loadRosterReadModel(cmd);
await enableFleetUnits(runner, roster, opts);
@@ -1758,37 +1609,6 @@ export function registerFleetCommand(program: Command, deps: FleetCommandDeps =
);
return;
}
if (action === 'start') {
// Broker preflight (#1292), re-probed on EVERY invocation: a
// gated runtime started without a live lease broker dies ~4s in
// while the unit reports active (RemainAfterExit) — enabling +
// starting here and then RE-CHECKING the socket refuses loudly
// instead of reporting rc0 over a doomed start. This is the
// second-start check as much as the first: it never trusts unit
// ActiveState.
await runChecked(runner, [
'systemctl',
'--user',
'enable',
'mosaic-lease-broker.service',
]);
await runChecked(runner, [
'systemctl',
'--user',
'start',
'mosaic-lease-broker.service',
]);
if (!(await brokerSocketPresent(deps))) {
console.error(
'[fleet] broker-absent: lease broker socket did not appear after enable+start (#1292).',
);
console.error(
'[fleet] remedy: mosaic fleet install (it reconciles either enable convention)',
);
process.exitCode = 1;
return;
}
}
if (action === 'restart') {
// Serialize the holder+agents teardown/relaunch behind the restart lock
// so a re-entrant restart waits for clean shutdown before relaunching,
@@ -2547,11 +2367,7 @@ export function registerFleetAgentCommands(
});
}
async function installFleet(
cmd: Command,
frameworkRoot: string,
runner: CommandRunner,
): Promise<void> {
async function installFleet(cmd: Command, frameworkRoot: string): Promise<void> {
const activePaths = resolveFleetPaths(cmd.opts<{ mosaicHome: string }>().mosaicHome);
assertDefaultMosaicHomeForSystemd(activePaths.mosaicHome);
// Read model first: every file this function places is roster-independent, and
@@ -2603,40 +2419,18 @@ async function installFleet(
for (const toolPath of executableToolPaths) {
await chmod(toolPath, 0o755);
}
// Unit placement (#1292): every unit goes through placeUnitFile — never a
// bare copyFile — so a by-path-enable symlink at the destination is
// unlinked rather than written through (copy-through would silently
// overwrite the SEED template, measured 2026-08-17). The lease broker unit
// is placed here too: previously the install named three units and omitted
// the broker entirely, which is why no documented path ever enabled it.
const placedUnits = await Promise.all(
[
'mosaic-tmux-holder.service',
'[email protected]',
'[email protected]',
'mosaic-lease-broker.service',
].map((unit) =>
placeUnitFile(join(frameworkRoot, 'systemd', 'user', unit), activePaths.systemdUserDir, unit),
),
await copyFile(
join(frameworkRoot, 'systemd', 'user', 'mosaic-tmux-holder.service'),
join(activePaths.systemdUserDir, 'mosaic-tmux-holder.service'),
);
const reconciled = placedUnits.filter(
(result) => result.unlinkedDestinationSymlink || result.removedStaleWantsSymlink,
await copyFile(
join(frameworkRoot, 'systemd', 'user', '[email protected]'),
join(activePaths.systemdUserDir, '[email protected]'),
);
await copyFile(
join(frameworkRoot, 'systemd', 'user', '[email protected]'),
join(activePaths.systemdUserDir, '[email protected]'),
);
if (reconciled.length > 0) {
console.log(
`Reconciled ${reconciled.length} unit placement(s) from by-path enable residue: ${reconciled.map((r) => r.unit).join(', ')}`,
);
}
// systemd will not see a replaced unit file without a reload; do it once
// after all placements, before any enable call below. runCommand never
// rejects (it resolves exitCode 127 on spawn error), so a plain await with
// an exitCode check matches the rest of this file's systemctl handling.
const reloadResult = await runner(...splitCommand(['systemctl', '--user', 'daemon-reload']));
if (reloadResult.exitCode !== 0) {
process.stderr.write(
`Warning: systemctl --user daemon-reload after unit placement failed (non-systemd host?): ${reloadResult.stderr || reloadResult.stdout || 'non-zero exit'}\n`,
);
}
// On roster v2 the reconciler owns the generated env: `apply` writes it and
// `regen` rebuilds it, both from projectRosterV2AgentGeneratedEnv. Writing it
@@ -2833,45 +2627,6 @@ function splitCommand(command: string[]): [string, string[]] {
return [bin, args];
}
/**
* Lease-broker socket presence for the fleet-start preflight (#1292).
* Resolution precedence matches launch.ts's defaultLeaseBrokerSocket and
* start-agent-session.sh's broker_socket_path: explicit
* MOSAIC_LEASE_BROKER_SOCKET, else $XDG_RUNTIME_DIR/mosaic-lease/broker.sock,
* else /run/user/<uid>/mosaic-lease/broker.sock. Pure filesystem check — this
* deliberately does NOT consult systemd state: a unit can be active
* (RemainAfterExit) with no live socket, and the socket is the thing the
* gated runtime connects to. Injectable via deps for tests.
*/
export function resolveLeaseBrokerSocketForPreflight(
env: NodeJS.ProcessEnv = process.env,
uid: number = typeof process.getuid === 'function' ? process.getuid() : 0,
): string {
if (env['MOSAIC_LEASE_BROKER_SOCKET']) return env['MOSAIC_LEASE_BROKER_SOCKET'];
const runtimeDir = env['XDG_RUNTIME_DIR'] ?? `/run/user/${uid}`;
return join(runtimeDir, 'mosaic-lease', 'broker.sock');
}
export async function brokerSocketPresent(
deps: FleetCommandDeps,
env: NodeJS.ProcessEnv = process.env,
): Promise<boolean> {
const check = deps.checkBrokerSocket;
const socketPath = resolveLeaseBrokerSocketForPreflight(env);
if (check) return check(socketPath);
// S_IFSOCK (0xC000) is a file-TYPE constant, not an access() mode (0-7):
// access(path, S_IFSOCK) throws ERR_OUT_OF_RANGE on node >= 24 (measured on
// v24.18.0, #1297 review F1) and cannot succeed on any node — the old catch
// swallowed the throw, so this probe could NEVER return true and every
// un-seamed call reported the broker absent. stat() + isSocket() is the real
// check and matches the bash side's [ -S ].
try {
return (await stat(socketPath)).isSocket();
} catch {
return false;
}
}
/** All supported fleet profile names. */
export type FleetProfile =
| 'general'
@@ -205,12 +205,6 @@ export async function runLeaseEnforcementDoctorCheck(
message:
`Lease-enforcement hooks (${matchedMarkers.join(', ')}) are wired in ~/.claude/settings.json, but ${reasons.join(' and ')}. ` +
'Every gated tool call will fail closed and BRICK this agent (see #869). ' +
// #1292: one remedy, correct under BOTH enable conventions (by-path on
// the seed template, and copy-then-enable in the active dir). Written
// from the 2026-08-17 symlink measurement: `systemctl enable` by name
// does NOT rewrite an existing by-path wants-symlink, so teaching a
// manual systemctl line here could leave a host with two competing
// wants links. fleet install reconciles either shape.
'Remedy: run `mosaic fleet install` (it reconciles either enable convention), or remove the enforcement hooks from ~/.claude/settings.json.',
'Remediate by activating the lease-broker supervisor (systemd unit + socket) or by removing the enforcement hooks from ~/.claude/settings.json.',
};
}
@@ -1,4 +1,4 @@
import { describe, it, expect, beforeEach, afterEach, vi } from 'vitest';
import { describe, it, expect, beforeEach, afterEach } from 'vitest';
import {
chmodSync,
mkdtempSync,
@@ -16,7 +16,6 @@ import {
renderPeerReach,
readFleetCommsBlock,
resolveCommsBlock,
resolveFleetIdentity,
resolvePeerCommand,
renderToolsContractStatus,
} from './comms-onboarding.js';
@@ -62,68 +61,6 @@ describe('shared fleet roster v1 resolver', () => {
});
});
// stack#1380 verification unblock: the fleet's own roster-v2 tooling writes
// the v1 body plus a generation fence and seat lifecycle/launch envelopes.
// The parser tolerates exactly that envelope (validated, opaque to comms).
const V2_ROSTER = [
'version: 2',
'generation: 8',
'transport: tmux',
'tmux:',
' socket_name: mosaic-fleet',
'defaults:',
' working_directory: ~/.mosaic',
' runtime: claude',
'agents:',
' - name: orch-01',
' runtime: claude',
' class: orchestrator',
' model: opus',
' reasoning: high',
' lifecycle:',
' enabled: true',
' desired_state: running',
' launch:',
' yolo: false',
'',
].join('\n');
it('accepts the roster-v2 envelope (generation + lifecycle/launch) on the v1 body', () => {
const resolved = parseFleetRosterV1(V2_ROSTER, 'yaml');
expect(resolved.tmux.socketName).toBe('mosaic-fleet');
expect(resolved.agents[0]?.name).toBe('orch-01');
});
it('rejects a non-integer generation', () => {
expect(() =>
parseFleetRosterV1(V2_ROSTER.replace('generation: 8', 'generation: eight'), 'yaml'),
).toThrow(/generation must be a non-negative integer/);
});
it('rejects an invalid lifecycle desired_state', () => {
expect(() =>
parseFleetRosterV1(
V2_ROSTER.replace('desired_state: running', 'desired_state: paused'),
'yaml',
),
).toThrow(/desired_state must be running\|stopped/);
});
it('rejects unknown fields inside the lifecycle envelope', () => {
expect(() =>
parseFleetRosterV1(
V2_ROSTER.replace(' enabled: true', ' enabled: true\n surprise: 1'),
'yaml',
),
).toThrow(/lifecycle has unknown field/);
});
it('rejects a non-boolean launch.yolo', () => {
expect(() =>
parseFleetRosterV1(V2_ROSTER.replace('yolo: false', 'yolo: sometimes'), 'yaml'),
).toThrow(/launch\.yolo must be a boolean/);
});
it('rejects unknown fields instead of leniently constructing a second roster view', () => {
expect(() => parseFleetRosterV1(`${ROSTER}\nunknown: value\n`, 'yaml')).toThrow(
/unknown field/i,
@@ -556,11 +493,6 @@ describe('resolvePeerCommand', () => {
describe('readFleetCommsBlock — spawned-agent context', () => {
let home: string;
beforeEach(() => {
// Hermetic helper fallback (stack#1380): the resolver probes
// $HOME/.config/mosaic when mosaicHome itself carries no helper — point
// HOME at a sandbox parent so tests never see the real host install.
vi.stubEnv('HOME', mkdtempSync(join(tmpdir(), 'mosaic-homeless-')));
vi.stubEnv('MOSAIC_HOME', '');
home = mkdtempSync(join(tmpdir(), 'mosaic-comms-'));
mkdirSync(join(home, 'fleet'), { recursive: true });
mkdirSync(join(home, 'tools', 'tmux'), { recursive: true });
@@ -569,10 +501,7 @@ describe('readFleetCommsBlock — spawned-agent context', () => {
writeFileSync(helper, '#!/bin/sh\n');
chmodSync(helper, 0o755);
});
afterEach(() => {
vi.unstubAllEnvs();
rmSync(home, { recursive: true, force: true });
});
afterEach(() => rmSync(home, { recursive: true, force: true }));
it('uses the authoritative self host and global socket from the shared roster resolver', () => {
const result = readFleetCommsBlock(home, 'enhancer', 'process-host-must-not-win');
@@ -635,27 +564,23 @@ describe('readFleetCommsBlock — spawned-agent context', () => {
},
],
[
'symlink escaping the install home',
'symlink',
() => {
const helper = join(home, 'tools', 'tmux', 'agent-send.sh');
rmSync(helper);
const outside = mkdtempSync(join(tmpdir(), 'mosaic-helper-outside-'));
writeFileSync(join(outside, 'real-send.sh'), '#!/bin/sh\n', { mode: 0o755 });
symlinkSync(join(outside, 'real-send.sh'), helper);
writeFileSync(join(home, 'real-send.sh'), '#!/bin/sh\n');
symlinkSync(join(home, 'real-send.sh'), helper);
},
],
['non-executable', () => chmodSync(join(home, 'tools', 'tmux', 'agent-send.sh'), 0o644)],
])(
'fails closed for a %s helper with deterministic guidance (no forbidden remedy)',
(_case, mutate) => {
mutate();
const result = readFleetCommsBlock(home, 'enhancer', 'w-jarvis');
expect(result.ok).toBe(false);
expect(result.output).toBe('');
expect(result.error).not.toContain('--repair-tools'); // stack#1380 M5a
expect(result.error).toContain('no active context or session was rewritten');
},
);
])('fails closed for a %s helper with deterministic repair guidance', (_case, mutate) => {
mutate();
const result = readFleetCommsBlock(home, 'enhancer', 'w-jarvis');
expect(result.ok).toBe(false);
expect(result.output).toBe('');
expect(result.error).toContain('mosaic update --repair-tools');
expect(result.error).toContain('no active context or session was rewritten');
});
it('does not rewrite the roster while resolving context', () => {
const path = join(home, 'fleet', 'roster.yaml');
@@ -678,11 +603,11 @@ describe('renderToolsContractStatus — non-mutating install drift', () => {
});
afterEach(() => rmSync(home, { recursive: true, force: true }));
it('names operator-verified recovery instead of a forbidden remedy when installed TOOLS.md is missing', () => {
it('uses the supported repair command when installed TOOLS.md is missing', () => {
const status = renderToolsContractStatus(home);
expect(status).toContain('authorized operator');
expect(status).not.toContain('--repair-tools'); // stack#1380 M5a
expect(status).toContain('mosaic update --repair-tools');
expect(status).not.toContain('--reseed');
expect(status).toContain('authorized operator');
});
it('reports stale preserved content without rewriting it', () => {
@@ -691,8 +616,8 @@ describe('renderToolsContractStatus — non-mutating install drift', () => {
writeFileSync(path, stale);
const status = renderToolsContractStatus(home);
expect(status).toContain('fleet-comms-contract: 1');
expect(status).toContain('authorized operator');
expect(status).not.toContain('--repair-tools'); // stack#1380 M5a
expect(status).toContain('digest-qualified backup');
expect(status).toContain('mosaic update --repair-tools');
expect(status).toContain('active context was not rewritten');
expect(readFileSync(path, 'utf8')).toBe(stale);
});
@@ -723,7 +648,7 @@ describe('renderToolsContractStatus — non-mutating install drift', () => {
expect(renderToolsContractStatus(home)).not.toBe('');
});
it('reads an installed TOOLS.md symlink whose validated target diverges (stack#1380 resolve-then-validate)', () => {
it('treats installed TOOLS.md symlinks as stale without following or rewriting them', () => {
const external = join(home, 'external-tools.md');
const externalContent = '# external\n<!-- fleet-comms-contract: 1 -->\n';
writeFileSync(external, externalContent);
@@ -731,23 +656,24 @@ describe('renderToolsContractStatus — non-mutating install drift', () => {
const status = renderToolsContractStatus(home);
expect(status).toContain('does not byte-match');
expect(status).toContain('unavailable');
expect(status).toContain('mosaic update --repair-tools');
expect(readFileSync(external, 'utf8')).toBe(externalContent);
});
it('treats a source TOOLS.md symlink escaping the install home as unavailable', () => {
const outside = mkdtempSync(join(tmpdir(), 'mosaic-source-outside-'));
it('treats source TOOLS.md symlinks as unavailable without following them', () => {
const external = join(home, 'external-source.md');
const content = '# authoritative tools\n<!-- fleet-comms-contract: 1 -->\n';
writeFileSync(join(outside, 'external-source.md'), content);
writeFileSync(external, content);
rmSync(join(home, 'defaults', 'TOOLS.md'));
symlinkSync(join(outside, 'external-source.md'), join(home, 'defaults', 'TOOLS.md'));
symlinkSync(external, join(home, 'defaults', 'TOOLS.md'));
writeFileSync(join(home, 'TOOLS.md'), content);
const status = renderToolsContractStatus(home);
expect(status).toContain('source contract');
expect(status).toContain('unavailable');
expect(readFileSync(join(outside, 'external-source.md'), 'utf8')).toBe(content);
expect(readFileSync(external, 'utf8')).toBe(content);
});
it('accepts byte-equal bounded source and installed contracts', () => {
@@ -804,91 +730,3 @@ describe('resolveCommsBlock — mosaic agent comms-block', () => {
expect(result.error).toContain('requires');
});
});
describe('resolveFleetIdentity — stack#1380 split-home layouts', () => {
// Brain-shaped mosaicHome (fleet state, NO tools/tmux) + framework config
// home carrying the helper, roster unified by the framework-created symlink
// <configHome>/fleet/roster.yaml -> <brain>/fleet/roster.yaml. This is the
// host layout that was down; all probes are POSITIONAL per the #1380
// verification protocol (an object arg proves nothing — M5b). HOME is
// stubbed so the config-default fallback stays inside the sandbox.
let brain: string;
let configHome: string;
beforeEach(() => {
const parent = mkdtempSync(join(tmpdir(), 'mosaic-i1380-parent-'));
brain = join(parent, 'brain');
// Framework home at the stubbed DEFAULT location so the fallback derives
// exactly as in production ($HOME/.config/mosaic), not by coincidence.
configHome = join(parent, 'home', '.config', 'mosaic');
vi.stubEnv('HOME', join(parent, 'home'));
vi.stubEnv('MOSAIC_HOME', '');
mkdirSync(join(brain, 'fleet'), { recursive: true });
writeFileSync(join(brain, 'fleet', 'roster.yaml'), ROSTER, { mode: 0o600 });
mkdirSync(join(configHome, 'fleet'), { recursive: true });
symlinkSync(join(brain, 'fleet', 'roster.yaml'), join(configHome, 'fleet', 'roster.yaml'));
mkdirSync(join(configHome, 'tools', 'tmux'), { recursive: true });
writeFileSync(join(configHome, 'tools', 'tmux', 'agent-send.sh'), '#!/bin/sh\n', {
mode: 0o755,
});
process.env['MOSAIC_BRAIN_HOME'] = brain;
});
afterEach(() => {
delete process.env['MOSAIC_BRAIN_HOME'];
vi.unstubAllEnvs();
rmSync(join(brain, '..'), { recursive: true, force: true });
});
it('resolves a member through the roster symlink under the config home', () => {
const result = resolveFleetIdentity(configHome, 'orchestrator', 'w-jarvis');
expect(result.ok).toBe(true);
expect(result.identity?.member.name).toBe('orchestrator');
expect(result.identity?.agentSendPath).toBe(join(configHome, 'tools', 'tmux', 'agent-send.sh'));
});
it('resolves a member when mosaicHome is the brain (helper found under the framework home)', () => {
const result = resolveFleetIdentity(brain, 'enhancer', 'w-jarvis');
expect(result.ok).toBe(true);
expect(result.identity?.member.name).toBe('enhancer');
expect(result.identity?.agentSendPath).toBe(join(configHome, 'tools', 'tmux', 'agent-send.sh'));
});
it('no-name control stays a quiet no-op', () => {
expect(resolveFleetIdentity(configHome, undefined, 'w-jarvis')).toEqual({ ok: true });
expect(resolveFleetIdentity(brain, undefined, 'w-jarvis')).toEqual({ ok: true });
});
it('a nonce name fails naming membership, not the symlink or the helper', () => {
const result = resolveFleetIdentity(configHome, 'nonce-' + Date.now(), 'w-jarvis');
expect(result.ok).toBe(false);
expect(result.error).not.toContain('symbolic link');
expect(result.error).not.toContain('helper');
expect(result.error).toContain('nonce-');
});
it('a non-member failure names membership, not the symlink (protocol control)', () => {
const result = resolveFleetIdentity(configHome, 'jarvis', 'w-jarvis');
expect(result.ok).toBe(false);
expect(result.error).not.toContain('symbolic link');
expect(result.error).not.toContain('helper is unavailable');
expect(result.error).toContain('orchestrator'); // known-member listing
});
it('names every searched framework home when the helper is missing everywhere', () => {
rmSync(join(configHome, 'tools'), { recursive: true, force: true });
const result = resolveFleetIdentity(brain, 'orchestrator', 'w-jarvis');
expect(result.ok).toBe(false);
expect(result.error).toContain('agent-send.sh');
expect(result.error).toContain(join(brain, 'tools', 'tmux', 'agent-send.sh'));
expect(result.error).not.toContain('--repair-tools');
});
it('readFleetCommsBlock composes the full contract on the split-home layout', () => {
const result = readFleetCommsBlock(configHome, 'orchestrator', 'w-jarvis');
expect(result.ok).toBe(true);
expect(result.output).toContain(
'Helper: `' + join(configHome, 'tools', 'tmux', 'agent-send.sh'),
);
});
});
+8 -66
View File
@@ -9,9 +9,8 @@
import { createHash } from 'node:crypto';
import { existsSync } from 'node:fs';
import { homedir, hostname } from 'node:os';
import { join, resolve } from 'node:path';
import { join } from 'node:path';
import { readRegularFileSecure } from './secure-file.js';
import { resolveBrainHome } from './brain-home.js';
import {
parseFleetRosterV1,
resolveInstalledFleetRosterPath,
@@ -261,13 +260,7 @@ context and have an authorized operator relaunch only this exact roster member w
function validateAgentSendHelper(path: string, mosaicHome: string): string | undefined {
try {
readRegularFileSecure(path, {
root: mosaicHome,
executable: true,
// The helper tree may live under the framework config home while this
// caller's mosaicHome is the brain; both are framework-owned roots.
symlinkTargetRoots: [resolveBrainHome(mosaicHome)],
});
readRegularFileSecure(path, { root: mosaicHome, executable: true });
return undefined;
} catch (error) {
const reason = error instanceof Error ? error.message : String(error);
@@ -275,48 +268,8 @@ function validateAgentSendHelper(path: string, mosaicHome: string): string | und
}
}
/**
* Framework install homes probed for tools/tmux/agent-send.sh (stack#1380 M2).
* The helper ships with the FRAMEWORK install, which on split-home layouts is
* the config home not the brain (~/.mosaic carries fleet state, no tools).
*/
function frameworkHelperHomes(mosaicHome: string): string[] {
const homes = [resolve(mosaicHome)];
const envHome = process.env['MOSAIC_HOME'];
if (envHome && envHome.trim() !== '' && resolve(envHome) !== resolve(mosaicHome)) {
homes.push(resolve(envHome));
}
const configDefault = join(homedir(), '.config', 'mosaic');
if (resolve(configDefault) !== resolve(mosaicHome)) homes.push(configDefault);
return homes;
}
function resolveAgentSendHelper(mosaicHome: string): { path: string; error?: string } {
const homes = frameworkHelperHomes(mosaicHome);
for (const home of homes) {
const helper = join(home, 'tools', 'tmux', 'agent-send.sh');
if (!existsSync(helper)) continue;
const error = validateAgentSendHelper(helper, home);
if (!error) return { path: helper };
// Present but unsafe: surface that verdict instead of silently probing on.
return { path: helper, error };
}
return {
path: join(resolve(mosaicHome), 'tools', 'tmux', 'agent-send.sh'),
error:
`fleet helper agent-send.sh was not found under any framework install home ` +
`(${homes.map((h) => join(h, 'tools', 'tmux', 'agent-send.sh')).join('; ')}). ` +
`Verify the framework install for this host (the helper ships with the framework ` +
`config home; the brain home carries fleet state, not tools) and have an authorized ` +
`operator restore it if missing.`,
};
}
function helperFailureGuidance(reason: string): string {
// stack#1380 M5a: `mosaic update --repair-tools` is a forbidden remedy on the
// affected estate (and wrong for a layout/missing-helper failure). Name the
// actual recovery shape instead.
return `${reason}. Verify the framework install provides tools/tmux/agent-send.sh under the framework config home and that the roster resolves (split-home layouts symlink the roster into the brain); contact the operator if it persists; no active context or session was rewritten.`;
return `${reason}. Run \`mosaic update --repair-tools\` to restore the supported current-version helper and TOOLS contract, then retry exact-member composition; no active context or session was rewritten.`;
}
export function resolveFleetIdentity(
@@ -325,15 +278,9 @@ export function resolveFleetIdentity(
localHost: string = shortHostname(),
): FleetIdentityResult {
if (!requestedName) return { ok: true };
const helper = resolveAgentSendHelper(mosaicHome);
if (helper.error) return { ok: false, error: helperFailureGuidance(helper.error) };
const agentSendPath = helper.path;
// Split-home layouts unify the roster by symlinking
// <configHome>/fleet/roster.yaml -> <brain>/fleet/roster.yaml. The secure
// read resolves that framework-created symlink when the brain is a
// sanctioned target root (stack#1380 M1).
const rosterSymlinkRoots = [resolveBrainHome(mosaicHome)];
const agentSendPath = join(mosaicHome, 'tools', 'tmux', 'agent-send.sh');
const helperError = validateAgentSendHelper(agentSendPath, mosaicHome);
if (helperError) return { ok: false, error: helperFailureGuidance(helperError) };
let rosterPath: string;
try {
@@ -354,10 +301,7 @@ export function resolveFleetIdentity(
let roster: FleetRoster;
try {
roster = parseFleetRosterV1(
readRegularFileSecure(rosterPath, {
root: mosaicHome,
symlinkTargetRoots: rosterSymlinkRoots,
}).content.toString('utf8'),
readRegularFileSecure(rosterPath, { root: mosaicHome }).content.toString('utf8'),
rosterPath.endsWith('.json') ? 'json' : 'yaml',
);
} catch (error) {
@@ -455,9 +399,7 @@ function boundedContractDigest(
}
function replacementGuidance(): string {
// stack#1380 M5a: never recommend the forbidden --repair-tools remedy from
// error text; name the operator-verified recovery shape instead.
return `Verify the installed TOOLS contract against the framework source with an authorized operator (the installed file must byte-match the supported current version) and have the operator explicitly relaunch the exact roster member. The active context was not rewritten.`;
return `Run \`mosaic update --repair-tools\` to make a digest-qualified backup and restore the supported current-version TOOLS contract, then have an authorized operator explicitly relaunch the exact roster member. The active context was not rewritten.`;
}
/** Detect preserved installed TOOLS.md drift without changing it. */
@@ -169,16 +169,6 @@ function reconcileDeps(host: FakeLifecycleHost): FleetReconcileDeps {
applyProjection: async () => undefined,
readRoster: async () => host.roster,
acquireMutationLock: async () => async () => undefined,
// Hermetic broker observation (#1297 F3): without this, the plan probes
// the REAL host filesystem, so the "stable JSON" fixtures answered true
// on any machine with a live lease broker and false elsewhere. Pointing
// both paths at fixtures that do not exist pins socketPresent:false and
// unitInstalled:false on every host, which is what these fixtures assert.
brokerSocketEnv: {
MOSAIC_LEASE_BROKER_SOCKET: '/nonexistent/mosaic-lease/broker.sock',
XDG_CONFIG_HOME: '/nonexistent/mosaic-config',
XDG_RUNTIME_DIR: '/nonexistent/run',
},
};
}
@@ -469,7 +459,6 @@ describe('FCM-M3-002 reconciler lifecycle acceptance', (): void => {
plan: {
generation: 7,
holder: 'owned',
broker: { unitInstalled: false, socketPresent: false },
agents: [
{
name: 'coder0',
@@ -1,7 +1,6 @@
import { chmod, mkdir, mkdtemp, readFile, rm, symlink, writeFile } from 'node:fs/promises';
import { tmpdir } from 'node:os';
import { join } from 'node:path';
import { createServer } from 'node:net';
import { afterEach, describe, expect, it } from 'vitest';
import {
acquirePrivateReconcileLock,
@@ -93,179 +92,6 @@ async function run(command: FleetReconcileCommand, overrides: Partial<FleetRecon
}
describe('fleet roster-owned reconciler', (): void => {
// ── #1292: broker as first-class plan member + broker-first start ordering ──
it('reports broker unit and socket state in the plan (socket is the signal, not unit state)', async (): Promise<void> => {
const result = await run('status', {
statPath: async () => true,
checkBrokerSocket: async () => true,
});
expect(result.plan.broker).toEqual({ unitInstalled: true, socketPresent: true });
});
it('reports a dead broker as socketPresent=false even when the unit is installed (enabled-but-dead is the #1292 shape)', async (): Promise<void> => {
const result = await run('status', {
statPath: async () => true,
checkBrokerSocket: async () => false,
});
expect(result.plan.broker).toEqual({ unitInstalled: true, socketPresent: false });
});
it('probes the REAL filesystem when no seam is injected — live socket and unit report healthy, absent paths report absent (#1297 F3)', async (): Promise<void> => {
const dir = await mkdtemp(join(tmpdir(), 'mosaic-broker-probe-'));
cleanup = dir;
const configHome = join(dir, 'config');
const unitDir = join(configHome, 'systemd', 'user');
await mkdir(unitDir, { recursive: true });
await writeFile(join(unitDir, 'mosaic-lease-broker.service'), '[Unit]\n');
const sockPath = join(dir, 'broker.sock');
const server = createServer();
await new Promise<void>((resolve) => {
server.listen(sockPath, resolve);
});
try {
const result = await run('status', {
brokerSocketEnv: {
MOSAIC_LEASE_BROKER_SOCKET: sockPath,
XDG_CONFIG_HOME: configHome,
XDG_RUNTIME_DIR: dir,
},
});
expect(result.plan.broker).toEqual({ unitInstalled: true, socketPresent: true });
// Absent paths through the SAME seam-less path answer false — this is
// the half the old default got right; healthy is the half it got wrong.
const absent = await run('status', {
brokerSocketEnv: {
MOSAIC_LEASE_BROKER_SOCKET: join(dir, 'gone.sock'),
XDG_CONFIG_HOME: join(dir, 'gone-config'),
},
});
expect(absent.plan.broker).toEqual({ unitInstalled: false, socketPresent: false });
} finally {
await new Promise<void>((resolve) => {
server.close(() => resolve());
});
}
});
it('command start refuses with a named error when the broker socket does not appear after enable+start (#1297 F3)', async (): Promise<void> => {
const calls: string[][] = [];
await expect(
run('start', {
checkBrokerSocket: async () => false,
runner: async (command, args) => {
calls.push([command, ...args]);
if (command === 'tmux' && args.includes('list-sessions')) {
return { stdout: '_holder\ncoder0\n', stderr: '', exitCode: 0 };
}
if (command === 'tmux' && args.includes('show-environment')) {
return {
stdout:
'HOME=/home/mosaic\nMOSAIC_FLEET_OWNER=11111111-1111-4111-8111-111111111111\nMOSAIC_TMUX_HOLDER=_holder\nMOSAIC_TMUX_SOCKET=mosaic-fleet\nPATH=/usr/bin:/bin\nPWD=/home/mosaic\n',
stderr: '',
exitCode: 0,
};
}
return { stdout: '', stderr: '', exitCode: 0 };
},
}),
).rejects.toThrow(/broker-absent/);
// Refused: broker enable+start attempted, no holder/agent unit touched.
const agentStarts = calls.filter(
(c) => c.join(' ') === 'systemctl --user start [email protected]',
);
expect(agentStarts).toHaveLength(0);
});
it('command start enables and starts the broker BEFORE the holder and any agent unit', async (): Promise<void> => {
const calls: string[][] = [];
const result = await run('start', {
// Deterministic broker presence: without the seam this test answers the
// HOST's broker state (passes on a machine with a live broker, refuses
// on CI), not the ordering property it exists for (#1297 follow-up).
checkBrokerSocket: async () => true,
runner: async (command, args) => {
calls.push([command, ...args]);
if (command === 'tmux' && args.includes('list-sessions')) {
return { stdout: '_holder\ncoder0\n', stderr: '', exitCode: 0 };
}
if (command === 'tmux' && args.includes('show-environment')) {
return {
stdout:
'HOME=/home/mosaic\nMOSAIC_FLEET_OWNER=11111111-1111-4111-8111-111111111111\nMOSAIC_TMUX_HOLDER=_holder\nMOSAIC_TMUX_SOCKET=mosaic-fleet\nPATH=/usr/bin:/bin\nPWD=/home/mosaic\n',
stderr: '',
exitCode: 0,
};
}
return { stdout: '', stderr: '', exitCode: 0 };
},
});
expect(result.lifecycle).toBe('complete');
const brokerEnable = calls.findIndex(
(c) => c.join(' ') === 'systemctl --user enable mosaic-lease-broker.service',
);
const brokerStart = calls.findIndex(
(c) => c.join(' ') === 'systemctl --user start mosaic-lease-broker.service',
);
const holderStart = calls.findIndex(
(c) => c.join(' ') === 'systemctl --user start mosaic-tmux-holder.service',
);
const agentStart = calls.findIndex(
(c) => c.join(' ') === 'systemctl --user start [email protected]',
);
expect(brokerEnable).toBeGreaterThanOrEqual(0);
expect(brokerStart).toBeGreaterThan(brokerEnable);
// Holder start may be absent (holder 'owned' in this fixture); if present it must follow the broker.
if (holderStart >= 0) expect(holderStart).toBeGreaterThan(brokerStart);
expect(agentStart).toBeGreaterThan(brokerStart);
});
it('apply with a running desired agent also enables and starts the broker first', async (): Promise<void> => {
const calls: string[][] = [];
const runningRoster: FleetRosterV2 = {
...roster,
agents: roster.agents.map((agent) =>
agent.name === 'coder0'
? { ...agent, lifecycle: { enabled: true, desiredState: 'running' as const } }
: agent,
),
};
const result = await executeFleetReconcile({
roster: runningRoster,
command: 'apply',
expectedGeneration: 7,
deps: deps({
readRoster: async () => runningRoster,
// Deterministic broker presence (see start-ordering test note).
checkBrokerSocket: async () => true,
runner: async (command, args) => {
calls.push([command, ...args]);
if (command === 'tmux' && args.includes('list-sessions')) {
return { stdout: '_holder\n', stderr: '', exitCode: 0 };
}
if (command === 'tmux' && args.includes('show-environment')) {
return {
stdout:
'HOME=/home/mosaic\nMOSAIC_FLEET_OWNER=11111111-1111-4111-8111-111111111111\nMOSAIC_TMUX_HOLDER=_holder\nMOSAIC_TMUX_SOCKET=mosaic-fleet\nPATH=/usr/bin:/bin\nPWD=/home/mosaic\n',
stderr: '',
exitCode: 0,
};
}
return { stdout: '', stderr: '', exitCode: 0 };
},
}),
});
expect(result.applied).toBe(true);
const brokerStart = calls.findIndex(
(c) => c.join(' ') === 'systemctl --user start mosaic-lease-broker.service',
);
const agentStart = calls.findIndex(
(c) => c.join(' ') === 'systemctl --user start [email protected]',
);
expect(brokerStart).toBeGreaterThanOrEqual(0);
expect(agentStart).toBeGreaterThan(brokerStart);
});
it('fails closed on a symlinked fleet ancestor without touching its target', async (): Promise<void> => {
const home = await lockHome();
const fleet = join(home, 'fleet');
@@ -549,8 +375,6 @@ describe('fleet roster-owned reconciler', (): void => {
expectedGeneration: 7,
deps: deps({
readRoster: async () => runningRoster,
// Deterministic broker presence (see start-ordering test note).
checkBrokerSocket: async () => true,
runner: async (command, args) => {
calls.push([command, ...args]);
if (command === 'tmux' && args.includes('list-sessions')) {
+13 -149
View File
@@ -1,5 +1,5 @@
import { constants } from 'node:fs';
import { lstat, open, readFile, stat, unlink, type FileHandle } from 'node:fs/promises';
import { lstat, open, readFile, unlink, type FileHandle } from 'node:fs/promises';
import { randomUUID } from 'node:crypto';
import { homedir } from 'node:os';
import { join } from 'node:path';
@@ -44,10 +44,6 @@ export interface FleetReconcileDeps {
readonly overrideDir?: string;
readonly homeDirectory?: string;
readonly readHolderIdentity?: () => Promise<string>;
/** Test/observation seams for the lease-broker plan member (#1292). */
readonly statPath?: (path: string) => Promise<boolean> | boolean;
readonly checkBrokerSocket?: (path: string) => Promise<boolean> | boolean;
readonly brokerSocketEnv?: NodeJS.ProcessEnv;
readonly validateRoster?: (roster: FleetRosterV2) => Promise<void>;
readonly prepareProjections?: (roster: FleetRosterV2) => Promise<readonly unknown[]>;
readonly applyProjection?: (prepared: unknown) => Promise<unknown>;
@@ -79,17 +75,6 @@ export interface FleetReconcileObservedAgent {
export interface FleetReconcilePlan {
readonly generation: number;
readonly holder: 'owned' | 'missing' | 'ownership-mismatch';
/**
* Lease broker observation (#1292): every gated runtime registers with the
* broker or dies ~4s in a broker not in the plan cannot be reported as
* drifted, which made "broker died an hour ago" and "broker fine"
* produce identical output. `unitInstalled` = unit file present in the
* active dir; `socketPresent` = live broker at the resolved socket path.
*/
readonly broker: {
readonly unitInstalled: boolean;
readonly socketPresent: boolean;
};
readonly agents: readonly FleetReconcileObservedAgent[];
readonly unmanagedSessions: readonly string[];
}
@@ -261,17 +246,7 @@ export async function executeFleetReconcile(
lifecycle: 'complete',
plan,
};
} catch (error: unknown) {
// A named lifecycle precondition (broker-absent after enable+start,
// #1297 F3) must surface as itself — converting it to the generic
// recoverable result would hide the diagnosis and report a clean
// refusal where a loud one is the point.
if (
error instanceof FleetReconcileError &&
error.code === 'lifecycle-precondition-failed'
) {
throw error;
}
} catch {
result = {
applied: false,
authoritativeRoster: 'unchanged',
@@ -340,63 +315,6 @@ function isObservational(command: FleetReconcileCommand): boolean {
return command === 'plan' || command === 'status' || command === 'verify' || command === 'doctor';
}
/**
* Observe the lease broker for the plan (#1292). Unit presence via systemctl
* is-system-running is NOT the signal a unit can be enabled-but-dead. The
* authoritative signal is the socket the gated runtimes connect to, matching
* broker-supervisor.ts's `checkBrokerSupervisorHealth` (healthy ===
* socketPresent). Injectable so tests drive every branch without a broker.
*/
function resolveBrokerSocketPath(env: NodeJS.ProcessEnv): string {
const uid = typeof process.getuid === 'function' ? process.getuid() : 0;
const runtimeDir = env['XDG_RUNTIME_DIR'] ?? `/run/user/${uid}`;
return env['MOSAIC_LEASE_BROKER_SOCKET'] ?? join(runtimeDir, 'mosaic-lease', 'broker.sock');
}
/**
* Probe the broker socket. Seams take precedence, but with no seam injected
* the REAL stat().isSocket() runs (#1297 review F3): production passes no
* seams, and defaulting to false made plan/status/doctor report a healthy
* broker as absent a dead broker was indistinguishable from noise.
*/
async function brokerSocketPresent(
deps: FleetReconcileDeps,
env: NodeJS.ProcessEnv,
): Promise<boolean> {
const socketPath = resolveBrokerSocketPath(env);
const check = deps.checkBrokerSocket;
if (check) return check(socketPath);
try {
return (await stat(socketPath)).isSocket();
} catch {
return false;
}
}
async function observeBroker(deps: FleetReconcileDeps): Promise<FleetReconcilePlan['broker']> {
const homeDirectory = deps.homeDirectory ?? homedir();
const env = (deps.brokerSocketEnv ?? process.env) as NodeJS.ProcessEnv;
const configHome = env['XDG_CONFIG_HOME'] ?? join(homeDirectory, '.config');
const unitPath = join(configHome, 'systemd', 'user', 'mosaic-lease-broker.service');
const statPath = deps.statPath;
let unitInstalled = false;
let socketPresent = false;
try {
// Same principle as the socket probe: no seam → look at the real
// filesystem. A unit file placed by installFleet (or a by-path residue
// symlink resolving to it) satisfies stat().isFile().
unitInstalled = statPath ? await statPath(unitPath) : (await stat(unitPath)).isFile();
} catch {
unitInstalled = false;
}
try {
socketPresent = await brokerSocketPresent(deps, env);
} catch {
socketPresent = false;
}
return { unitInstalled, socketPresent };
}
async function observeFleet(
roster: FleetRosterV2,
deps: FleetReconcileDeps,
@@ -407,12 +325,10 @@ async function observeFleet(
'-F',
'#{session_name}',
]);
const broker = await observeBroker(deps);
if (sessionsResult.exitCode !== 0) {
return {
generation: roster.generation,
holder: 'missing',
broker,
agents: await observeAgents(roster, deps, new Set<string>()),
unmanagedSessions: [],
};
@@ -435,7 +351,6 @@ async function observeFleet(
return {
generation: roster.generation,
holder,
broker,
agents: await observeAgents(roster, deps, sessions),
unmanagedSessions: Object.freeze(unmanagedSessions.sort()),
};
@@ -592,17 +507,6 @@ async function executeExplicitLifecycle(
plan: FleetReconcilePlan,
agents: readonly FleetRosterV2Agent[],
): Promise<FleetReconcileResult> {
const lifecycleApplyFailed = (): FleetReconcileResult => ({
applied: false,
authoritativeRoster: 'unchanged',
projections: 'not-applied',
lifecycle: 'incomplete',
plan,
recovery: {
code: 'lifecycle-apply-failed',
action: 'rerun-after-inspecting-owned-resources',
},
});
if (request.command === 'start') {
for (const agent of agents) {
if (!agent.lifecycle.enabled) {
@@ -613,40 +517,6 @@ async function executeExplicitLifecycle(
}
}
}
// Broker FIRST (#1292): a gated runtime started without a running lease
// broker dies ~4 seconds in at registration — enable the unit (install
// places it) and start it before any holder/agent lifecycle effect.
try {
if (request.command === 'start') {
await runChecked(request.deps, 'systemctl', [
'--user',
'enable',
'mosaic-lease-broker.service',
]);
await runChecked(request.deps, 'systemctl', [
'--user',
'start',
'mosaic-lease-broker.service',
]);
}
} catch {
return lifecycleApplyFailed();
}
if (request.command === 'start') {
// Socket re-check after start, as a NAMED precondition (#1297 review
// F3) — the same protection the v1 path in commands/fleet.ts has had all
// along: the unit reporting active is not the signal; the socket is.
// Deliberately outside the try/catch above: a swallowed FleetReconcileError
// here read as a generic recoverable failure, hiding the named refusal.
// Runs BEFORE any holder/agent unit is touched so nothing doomed starts.
const env = (request.deps.brokerSocketEnv ?? process.env) as NodeJS.ProcessEnv;
if (!(await brokerSocketPresent(request.deps, env))) {
throw new FleetReconcileError(
'lifecycle-precondition-failed',
'broker-absent: lease broker socket did not appear after enable+start (#1292; #1297 F3). Remedy: mosaic fleet install.',
);
}
}
try {
if (request.command === 'start' && plan.holder === 'missing') {
await runChecked(request.deps, 'systemctl', [
@@ -663,7 +533,17 @@ async function executeExplicitLifecycle(
]);
}
} catch {
return lifecycleApplyFailed();
return {
applied: false,
authoritativeRoster: 'unchanged',
projections: 'not-applied',
lifecycle: 'incomplete',
plan,
recovery: {
code: 'lifecycle-apply-failed',
action: 'rerun-after-inspecting-owned-resources',
},
};
}
return {
applied: true,
@@ -683,22 +563,6 @@ async function applyDesiredLifecycle(
(agent: FleetRosterV2Agent): boolean =>
agent.lifecycle.enabled && agent.lifecycle.desiredState === 'running',
);
// Broker before any running agent, same ordering and reason as the
// command-driven path above (#1292).
if (needsRunningAgent) {
await runChecked(deps, 'systemctl', ['--user', 'enable', 'mosaic-lease-broker.service']);
await runChecked(deps, 'systemctl', ['--user', 'start', 'mosaic-lease-broker.service']);
// Same socket re-check as the explicit start path (#1297 F3): apply with
// running desired agents starts gated runtimes too, and a broker that
// starts but never binds dooms them the same way.
const env = (deps.brokerSocketEnv ?? process.env) as NodeJS.ProcessEnv;
if (!(await brokerSocketPresent(deps, env))) {
throw new FleetReconcileError(
'lifecycle-precondition-failed',
'broker-absent: lease broker socket did not appear after enable+start (#1292; #1297 F3). Remedy: mosaic fleet install.',
);
}
}
if (needsRunningAgent && plan.holder === 'missing') {
await runChecked(deps, 'systemctl', ['--user', 'start', 'mosaic-tmux-holder.service']);
}
+1 -65
View File
@@ -7,7 +7,6 @@ import { canonicalizeRoleClass } from '../commands/fleet-personas.js';
interface RawFleetRoster {
version?: unknown;
transport?: unknown;
generation?: unknown;
tmux?: {
socket_name?: unknown;
socketName?: unknown;
@@ -42,10 +41,6 @@ interface RawFleetRoster {
resetBetweenTasks?: unknown;
kickstart_template?: unknown;
kickstartTemplate?: unknown;
model?: unknown;
reasoning?: unknown;
lifecycle?: { enabled?: unknown; desired_state?: unknown };
launch?: { yolo?: unknown };
}>;
connector?: {
kind?: unknown;
@@ -221,17 +216,7 @@ function normalizeFleetRosterV1Unchecked(raw: RawFleetRoster): FleetRoster {
'runtimes',
'agents',
'connector',
// stack#1380 verification unblock: the fleet's own roster-v2 mutation
// tooling writes a `generation` fence on the same v1 body. Tolerated here
// as an opaque non-negative integer; comms semantics are unchanged.
'generation',
]);
if (
raw.generation !== undefined &&
(typeof raw.generation !== 'number' || !Number.isInteger(raw.generation) || raw.generation < 0)
) {
throw new Error('Fleet roster generation must be a non-negative integer.');
}
if (raw.tmux !== undefined) {
assertObject(raw.tmux, 'Fleet roster tmux');
assertKnownKeys(raw.tmux, 'Fleet roster tmux', [
@@ -246,8 +231,6 @@ function normalizeFleetRosterV1Unchecked(raw: RawFleetRoster): FleetRoster {
assertKnownKeys(raw.defaults, 'Fleet roster defaults', [
'working_directory',
'workingDirectory',
// stack#1380 verification unblock: roster-v2 default runtime hint.
'runtime',
]);
}
if (raw.runtimes !== undefined) {
@@ -260,9 +243,7 @@ function normalizeFleetRosterV1Unchecked(raw: RawFleetRoster): FleetRoster {
]);
}
}
if (raw.version !== 1 && raw.version !== 2) {
throw new Error('Fleet roster version must be 1 or 2.');
}
if (raw.version !== 1) throw new Error('Fleet roster version must be 1.');
if (raw.transport !== 'tmux') throw new Error('Fleet roster transport must be "tmux".');
if (!Array.isArray(raw.agents) || raw.agents.length === 0) {
throw new Error('Fleet roster must define at least one agent.');
@@ -337,52 +318,7 @@ function normalizeAgent(raw: NonNullable<RawFleetRoster['agents']>[number]): Fle
'resetBetweenTasks',
'kickstart_template',
'kickstartTemplate',
// stack#1380 verification unblock: roster-v2 envelope fields written by
// the fleet's own mutation tooling. Validated, then opaque to comms.
'model',
'reasoning',
'lifecycle',
'launch',
]);
if (raw.model !== undefined && typeof raw.model !== 'string') {
throw new Error('Fleet roster agent model must be a string.');
}
if (raw.reasoning !== undefined && typeof raw.reasoning !== 'string') {
throw new Error('Fleet roster agent reasoning must be a string.');
}
const lifecycle = raw.lifecycle as { enabled?: unknown; desired_state?: unknown } | undefined;
if (lifecycle !== undefined) {
if (typeof lifecycle !== 'object' || lifecycle === null) {
throw new Error('Fleet roster agent lifecycle must be an object.');
}
const lifecycleKeys = Object.keys(lifecycle);
if (!lifecycleKeys.every((key) => key === 'enabled' || key === 'desired_state')) {
throw new Error('Fleet roster agent lifecycle has unknown field(s).');
}
if (lifecycle.enabled !== undefined && typeof lifecycle.enabled !== 'boolean') {
throw new Error('Fleet roster agent lifecycle.enabled must be a boolean.');
}
if (
lifecycle.desired_state !== undefined &&
(typeof lifecycle.desired_state !== 'string' ||
!['running', 'stopped'].includes(lifecycle.desired_state))
) {
throw new Error('Fleet roster agent lifecycle.desired_state must be running|stopped.');
}
}
const launch = raw.launch as { yolo?: unknown } | undefined;
if (launch !== undefined) {
if (typeof launch !== 'object' || launch === null) {
throw new Error('Fleet roster agent launch must be an object.');
}
const launchKeys = Object.keys(launch);
if (!launchKeys.every((key) => key === 'yolo')) {
throw new Error('Fleet roster agent launch has unknown field(s).');
}
if (launch.yolo !== undefined && typeof launch.yolo !== 'boolean') {
throw new Error('Fleet roster agent launch.yolo must be a boolean.');
}
}
const name = stringValue(raw.name, '', 'Fleet roster agent name');
const runtime = stringValue(
raw.runtime,
+11 -81
View File
@@ -10,14 +10,12 @@ import {
type PathLike,
} from 'node:fs';
import type * as NodeFs from 'node:fs';
import type { Stats } from 'node:fs';
import { tmpdir } from 'node:os';
import { join, resolve } from 'node:path';
import { join } from 'node:path';
interface FilesystemRaceState {
afterLstat?: (path: string) => void;
afterOpen?: (path: string) => void;
afterStat?: (path: string, stats: Stats) => Stats;
}
const filesystemRaceState = vi.hoisted<FilesystemRaceState>(() => ({}));
@@ -31,10 +29,6 @@ vi.mock('node:fs', async (importOriginal) => {
filesystemRaceState.afterLstat?.(String(path));
return result;
},
statSync: (path: PathLike) => {
const result = actual.statSync(path);
return filesystemRaceState.afterStat?.(String(path), result) ?? result;
},
openSync: (path: PathLike, flags: string | number, mode?: number) => {
const fd = actual.openSync(path, flags, mode);
filesystemRaceState.afterOpen?.(String(path));
@@ -52,13 +46,11 @@ describe('secure file reads', () => {
root = mkdtempSync(join(tmpdir(), 'mosaic-secure-file-'));
filesystemRaceState.afterLstat = undefined;
filesystemRaceState.afterOpen = undefined;
filesystemRaceState.afterStat = undefined;
});
afterEach(() => {
filesystemRaceState.afterLstat = undefined;
filesystemRaceState.afterOpen = undefined;
filesystemRaceState.afterStat = undefined;
rmSync(root, { recursive: true, force: true });
});
@@ -68,87 +60,25 @@ describe('secure file reads', () => {
);
});
// stack#1380: the guard resolves symlinks and validates the resolved target
// instead of refusing any symlink component.
it('permits a symlink ancestor whose resolved target is inside the root', () => {
it('rejects a symlink in a file ancestor', () => {
const external = join(root, 'external');
mkdirSync(external);
writeFileSync(join(external, 'file'), 'external\n');
symlinkSync(external, join(root, 'linked'));
const snapshot = readRegularFileSecure(join(root, 'linked', 'file'), { root });
expect(snapshot.content.toString('utf8')).toBe('external\n');
});
it('permits a symlinked file whose resolved target is inside the root', () => {
const external = join(root, 'external-file');
writeFileSync(external, 'external\n');
symlinkSync(external, join(root, 'linked-file'));
const snapshot = readRegularFileSecure(join(root, 'linked-file'), { root });
expect(snapshot.content.toString('utf8')).toBe('external\n');
});
it('permits a symlink resolving into an additional sanctioned root (split-home roster shape)', () => {
const brain = `${root}-brain`;
mkdirSync(join(brain, 'fleet'), { recursive: true });
writeFileSync(join(brain, 'fleet', 'roster.yaml'), 'roster\n', { mode: 0o600 });
mkdirSync(join(root, 'fleet'));
symlinkSync(join(brain, 'fleet', 'roster.yaml'), join(root, 'fleet', 'roster.yaml'));
const snapshot = readRegularFileSecure(join(root, 'fleet', 'roster.yaml'), {
root,
symlinkTargetRoots: [brain],
});
expect(snapshot.content.toString('utf8')).toBe('roster\n');
});
it('refuses a symlink whose resolved target escapes every sanctioned root', () => {
const outside = mkdtempSync(join(tmpdir(), 'mosaic-secure-outside-'));
try {
mkdirSync(join(root, 'fleet'), { recursive: true });
writeFileSync(join(outside, 'roster.yaml'), 'escaped\n', { mode: 0o600 });
symlinkSync(join(outside, 'roster.yaml'), join(root, 'fleet', 'roster.yaml'));
expect(() => readRegularFileSecure(join(root, 'fleet', 'roster.yaml'), { root })).toThrow(
/symlink target escapes managed roots/,
);
} finally {
rmSync(outside, { recursive: true, force: true });
}
});
it('refuses a group-writable symlink target', () => {
const loose = join(root, 'loose');
mkdirSync(loose);
chmodSync(loose, 0o770); // group-writable bit survives umask via explicit chmod
writeFileSync(join(loose, 'file'), 'loose\n');
symlinkSync(loose, join(root, 'linked-loose'));
expect(() => readRegularFileSecure(join(root, 'linked-loose', 'file'), { root })).toThrow(
/group- or world-writable/,
expect(() => readRegularFileSecure(join(root, 'linked', 'file'), { root })).toThrow(
'path ancestor is a symbolic link',
);
});
it('refuses a symlink target owned by another user', () => {
const external = join(root, 'foreign');
mkdirSync(external);
writeFileSync(join(external, 'file'), 'foreign\n');
symlinkSync(external, join(root, 'linked-foreign'));
it('rejects a symlink target', () => {
const external = join(root, 'external');
writeFileSync(external, 'external\n');
symlinkSync(external, join(root, 'linked-file'));
filesystemRaceState.afterStat = (path, stats): Stats => {
if (resolve(path) === resolve(external)) {
return { ...stats, uid: stats.uid + 4242 } as Stats;
}
return stats;
};
try {
expect(() => readRegularFileSecure(join(root, 'linked-foreign', 'file'), { root })).toThrow(
/not owned by the current user/,
);
} finally {
filesystemRaceState.afterStat = undefined;
}
expect(() => readRegularFileSecure(join(root, 'linked-file'), { root })).toThrow(
'file is a symbolic link',
);
});
it('keeps ancestor traversal bound when an opened directory is substituted', () => {
+26 -100
View File
@@ -7,25 +7,14 @@ import {
mkdirSync,
openSync,
readFileSync,
readlinkSync,
statSync,
} from 'node:fs';
import { platform } from 'node:os';
import { basename, dirname, isAbsolute, relative, resolve, sep } from 'node:path';
import { dirname, isAbsolute, relative, resolve, sep } from 'node:path';
export interface SecureFileReadOptions {
root: string;
maxBytes?: number;
executable?: boolean;
/**
* Additional roots a symlink component may resolve into (stack#1380).
* Default: only the managed root itself. Every symlink hop is validated
* containment under the root or one of these roots, current-user ownership,
* no group/world-writable mode and refusal stays the default for anything
* else. Callers that operate the split-home layout pass the brain home so
* the framework-created roster symlink resolves.
*/
symlinkTargetRoots?: string[];
}
export interface SecureFileSnapshot {
@@ -99,57 +88,7 @@ function openDirectoryChain(absoluteDirectory: string): { fd: number; descriptor
}
}
function containedUnder(root: string, target: string): boolean {
const rel = relative(resolve(root), resolve(target));
return rel !== '..' && !rel.startsWith(`..${sep}`) && !isAbsolute(rel) && rel !== '';
}
const MAX_SYMLINK_HOPS = 40;
/**
* Resolve every symlink on `lexical` component-wise, validating each hop
* (stack#1380 resolve-then-validate): the hop target must stay under one of
* the sanctioned roots, must be owned by the current user (or root), and must
* not be group- or world-writable. Returns a symlink-free absolute path.
*/
function resolveRealPath(lexical: string, sanctionedRoots: string[]): string {
const hopTargets: string[] = [];
let current: string = sep;
for (const piece of resolve(lexical).split(sep).filter(Boolean)) {
current = resolve(current, piece);
for (let hops = 0; lstatSync(current).isSymbolicLink(); ) {
if (++hops > MAX_SYMLINK_HOPS) {
throw new Error(`symlink chain exceeds ${MAX_SYMLINK_HOPS} hops: ${lexical}`);
}
const linkTarget = readlinkSync(current);
const absolute = resolve(dirname(current), linkTarget);
if (!sanctionedRoots.some((root) => containedUnder(root, absolute))) {
throw new Error(
`symlink target escapes managed roots [${sanctionedRoots.join(', ')}]: ${absolute}`,
);
}
hopTargets.push(absolute);
current = absolute;
}
}
const uid = typeof process.getuid === 'function' ? process.getuid() : 0;
for (const hop of hopTargets) {
const stat = statSync(hop);
if (stat.uid !== uid && stat.uid !== 0) {
throw new Error(`symlink target is not owned by the current user: ${hop}`);
}
if (stat.mode & 0o022) {
throw new Error(`symlink target is group- or world-writable: ${hop}`);
}
}
return current;
}
function openFileBeneathRoot(
root: string,
target: string,
symlinkTargetRoots: string[] = [],
): { fd: number; descriptors: number[] } {
function openFileBeneathRoot(root: string, target: string): { fd: number; descriptors: number[] } {
const canonicalRoot = resolve(root);
const canonicalTarget = resolve(target);
assertCanonicalContainment(canonicalRoot, canonicalTarget);
@@ -157,52 +96,39 @@ function openFileBeneathRoot(
const fileName = components.pop();
if (fileName === undefined) throw new Error('managed file path names the managed root');
// stack#1380: resolve-then-validate. The lexical path must name the managed
// root (above); symlink components are then resolved hop-by-hop under the
// sanctioned roots (validated per hop), and the descriptor traversal walks
// the symlink-free real path — keeping the O_NOFOLLOW chain as the race
// guard for anything substituted after resolution.
let realRoot: string;
try {
realRoot = resolveRealPath(canonicalRoot, [canonicalRoot]);
} catch (error) {
throw secureFilesystemError(
'secure descriptor traversal failed: symbolic link, unavailable, or not a directory',
error,
);
}
const sanctioned = [realRoot, ...symlinkTargetRoots.map((extra) => resolve(extra))];
let realTarget: string;
try {
realTarget = resolveRealPath(canonicalTarget, sanctioned);
} catch (error) {
if (error instanceof Error && !('code' in error)) throw error;
throw secureFilesystemError(
'secure descriptor traversal failed: symbolic link, unavailable, or not a directory',
error,
);
}
if (!sanctioned.some((sr) => containedUnder(sr, realTarget) || resolve(sr) === realTarget)) {
throw new Error(
`resolved path escapes managed roots [${sanctioned.join(', ')}]: ${realTarget}`,
);
}
const chain = openDirectoryChain(dirname(realTarget));
const rootChain = openDirectoryChain(canonicalRoot);
try {
let parentFd = rootChain.fd;
for (const component of components) {
try {
parentFd = openSync(
procDescriptorPath(parentFd, component),
constants.O_RDONLY | constants.O_DIRECTORY | constants.O_NOFOLLOW,
);
} catch (error) {
throw secureFilesystemError(
'path ancestor is a symbolic link, unavailable, or not a directory',
error,
);
}
rootChain.descriptors.push(parentFd);
if (!fstatSync(parentFd).isDirectory()) {
throw new Error('path ancestor is a symbolic link or not a directory');
}
}
let fd: number;
try {
fd = openSync(
procDescriptorPath(chain.fd, basename(realTarget)),
procDescriptorPath(parentFd, fileName),
constants.O_RDONLY | constants.O_NONBLOCK | constants.O_NOFOLLOW,
);
} catch (error) {
throw secureFilesystemError('file is a symbolic link or unavailable', error);
}
chain.descriptors.push(fd);
return { fd, descriptors: chain.descriptors };
rootChain.descriptors.push(fd);
return { fd, descriptors: rootChain.descriptors };
} catch (error) {
closeDescriptors(chain.descriptors);
closeDescriptors(rootChain.descriptors);
if (error instanceof Error) throw error;
throw new Error('secure managed file open failed');
}
@@ -277,7 +203,7 @@ export function readRegularFileSecure(
path: string,
options: SecureFileReadOptions,
): SecureFileSnapshot {
const openedFile = openFileBeneathRoot(options.root, path, options.symlinkTargetRoots ?? []);
const openedFile = openFileBeneathRoot(options.root, path);
try {
const opened = fstatSync(openedFile.fd);
if (!opened.isFile()) throw new Error('managed file is not a regular file');
@@ -168,9 +168,7 @@ describe('repairFleetCommsTools', () => {
const result = repairFleetCommsTools(framework, home);
expect(result).toMatchObject({ ok: false, changed: false });
// stack#1380: resolve-then-validate — an escaping symlink is still
// refused, with the new escape diagnostic.
expect(result.reason).toContain('symlink target escapes managed roots');
expect(result.reason).toContain('symbolic link');
expect(readFileSync(target, 'utf8')).toBe('do not touch\n');
expect(lstatSync(join(home, 'tools', 'tmux', 'agent-send.sh')).isSymbolicLink()).toBe(true);
});
@@ -224,10 +222,7 @@ describe('repairFleetCommsTools', () => {
const result = repairFleetCommsTools(framework, targetHome);
expect(result, testCase.name).toMatchObject({ ok: false, changed: false });
// stack#1380: escaping ancestor symlinks stay refused. The home case is
// caught by the managed-root guard ('is a symbolic link'); deeper
// components by resolve-then-validate ('symlink target escapes').
expect(result.reason, testCase.name).toMatch(/symbolic link|symlink target escapes/);
expect(result.reason, testCase.name).toContain('symbolic link');
expect(readdirSync(external), testCase.name).toEqual([]);
}
});
-2
View File
@@ -61,8 +61,6 @@ export const STAGES = [
'bash packages/mosaic/framework/tools/quality/scripts/check-tools-index.sh --self-test',
'bash packages/mosaic/framework/tools/quality/scripts/check-tools-index.sh',
'bash packages/mosaic/framework/tools/git/test-issue-close-fail-closed.sh',
'bash packages/mosaic/framework/tools/git/test-gitea-login-resolution.sh',
'bash packages/mosaic/framework/tools/git/test-issue-view-comments.sh',
'bash packages/mosaic/framework/tools/git/test-wrapper-guard.sh',
'bash packages/mosaic/framework/tools/git/test-mosaic-worktree-large-repo.sh',
],