15 Commits
Author SHA1 Message Date
jason.woltje 87db40e0cd fix(cli,seat,discord,organization,control-board,harness,plugins,webui): review round: launcher reasons, session, discord reads, mosaic-config and keys by label (#1873)
Review "changes 8" fix round for org-redact6.

- launcher: a harness module's throw names its system code or class,
  never its message; preflight, loader, bundle, checkout, sandbox and
  route text is relayed with host paths replaced (relay()).
- claude-code and pi preflights name MOSAIC_CLAUDE_BIN and the
  repository's node_modules, never a path; persona and skill bundle
  errors and sandboxProblem name their file or the sandbox by label.
- seat session: thread store, legacy store, overlap, busy lock and
  newRun refusals name the place under the data root.
- discord: unknown keys, binding names and turn ids are described when
  not plain; approvals, journal and repository read errors name the file
  and code; git drops stderr lines that carry a host path and a spawn
  failure names its code only.
- seat, control-board: registration fields and seen keys are described
  when not plain; invalid seat name and layout state the constraint.
- organization: the registry names the config directory and registry
  file by label; --project is quoted only when it is a plain name;
  systemConfigProblem moves to its own module.
- scripts/mosaic-config.mjs: the config by where it came from, the data
  root and its marker by role, keys and arguments only when plain, and
  error codes in place of raw messages. Exit codes unchanged.
- webui sso-config: an unknown console key is named only when plain.

Sentinel tests in each package's path-redaction-labels file fail on
BASE; pins that quoted a path now assert the label and its absence.
2026-11-11 08:53:31 -06:00
jason.woltje 623388603b Merge organizations MVP (ui-org-r2) into refactor
BUILD-LOG.md: append-only union, refactor entries first, then the branch's.
docs/SESSIONS.md: append-only union, refactor entries first, then the branch's.
packages/bus/src/runtime.mjs: keep the organization usage timer and refactor's halted/invalid startup-recovery gate.
packages/harness/src/first-output.mjs: keep the branch's harness-neutral markers; refactor's empty-length rejection moves into the Pi monitor (packages/plugins/harness/pi/model-stream.mjs).
packages/harness/tests/runner.test.mjs: keep refactor's length-continuation cases and the branch's output-policy cases.
packages/plugins/harness/pi/index.mjs: keep the documented GLM limits and the branch's transport-based bundle.

Suites: 6356 pass, 32 fail; all 32 fail identically on both parents
(node_modules symlink EISDIR in pinned-copy and stack copy/release tests,
missing Python jsonschema in two conversation tests).
2026-11-08 21:38:42 -06:00
jason.woltje c09570d696 Add plugin-based organization quick setup 2026-11-07 21:43:03 -06:00
jason.woltje af9e367719 Keep fixture harness selection in its plugin and run pi offline 2026-11-07 12:39:33 -06:00
jason.woltje b48396b68e fix(plugins): provider review fixes, admission binding, alias-safe models, no provider names in core (#1858) 2026-11-06 19:52:57 -06:00
jason.woltje ccccac2ade feat(plugins): model providers as plugin modules, cliproxy catalogue on Routing (#1858) 2026-11-06 19:11:13 -06:00
jason.woltje 08b7db413b fix(harness): move harness-specific output handling into the harness plugins (#1855) 2026-11-06 18:27:58 -06:00
jason.woltje 03072cc08b fix(plugins): set documented output limit for glm-5.3-flash (#1852)
Register the documented 65536-token default output limit and retain the existing 128000-token context window for glm-5.3-flash. Other model registrations are unchanged.

Pi's isRecoverableLength (pi-ai dist/utils/overflow.js:163), used by agent-session.js:1659, now treats a length stop below 65536 output tokens as recoverable, including context-clamped requests or a lower provider cap. Pi then compacts and retries. It allows one recovery in a row: the flag resets on a new user message or a successful assistant message that is not a length stop, so tool rounds can re-arm it within one prompt.

The recovery happens inside one runner attempt. The turn marker reflects the final agent_end, and the failed and retried messages still count in that message's usage delta, so three runner attempts need not mean three provider calls.

The 128000 context and 16384 reserve are unchanged, so threshold compaction still starts above 111616 tokens. Output clamping now starts above about 58368 estimated input tokens instead of 107520. No new local refusal.

This commit can be dropped without affecting Part A.
2026-11-06 16:49:31 -06:00
jason.woltje 8094901bce fix(harness): preserve failed Claude result diagnostics (#797) 2026-11-05 21:17:18 -06:00
jason.woltje 1bd0077391 fix(harness): mark model activity before final answers (#797) 2026-11-05 20:45:03 -06:00
jason.woltje 39eadf8ba8 Row 332 R2 (#1830): hermetic pi launches in cli tests, pi bundle default, review F2 and F3
The cli tests that launch pi deleted only the session login names, so pi's
preflight passed or refused by whatever provider keys the host exported. A
shared helper now deletes every credential pi's preflight reads and gives pi
launches one made-up key. The refusal itself is unchanged.

pi bundle() takes a default argument again, so a call with none returns the
non-proxy bundle, as before P3. pi.bundle leaves out a launch.models entry
whose provider has no proxy route instead of failing every pi launch (F2);
malformed proxy entries still refuse. The invalid session proxy context
BundleError gets the code invalid-proxy-context. New tests cover F2, the code,
and the three proxy account branches in the launcher (F3).
2026-11-02 15:22:17 -06:00
jason.woltje d6dc04e84f Join P3 harness context to landed P2 admission with private same-key delivery 2026-10-26 18:02:58 -05:00
jason.woltje dbbda25373 gizmoduck: row 307 (#1814, Vikunja 598) harness kind gains compactions: pi's stream scan and Claude-history check move to packages/plugins/harness/pi/compactions.mjs, claude-code answers harness-unsupported, core keeps the ledger and checks each answer, readLedger reads the reason only, the launcher passes the loaded module; four old-test edits under mos-1459 2026-10-23 09:37:50 -05:00
jason.woltje 2c38325eca row 66: managed sessions run the unslop check on both harnesses, the operator note on the RESULT, a correction notice on the next turn, and unslop.jsonl counts (#1559)
Lands Sage's sage/row66-r5 at 7ffb640ac, round 4 (r4d 82c830779) rebased onto a36523713 after row 197,
approved by Dewey (comment 30085) and Filbert (30086) on manifest 6cfd812da4ca (23 paths, rev 2905). The
rebase had no conflict. The range-diff is = except one pi-session.test.mjs hunk header. Patch-id without
agents/ is d67275ac4251 on 7b9737953, a36523713 and this commit's parent. Gate (mos-1186): unslop,
harness, cli and every scripts/test-*.sh green on r4d, harness 149/0 again on r5. Landing sessions on a
scratch root from the candidate tree: Claude Code r4ab24630d323 showed the note, the injected notice and
counts checked 2, flagged 1, notices 1. Pi r1dd94171bede is launch only until #383 (Vikunja #454).
2026-10-21 13:51:28 -05:00
jason.woltje c216f47dbb plugins, cli, bus, business, webui: harness, model and thinking from the instance vars, a fallback route, and a harness plug-in kind (row 65, #1558)
Adds the harness plug-in kind (interface 1). packages/plugins/harness/pi
and claude-code hold what core branched on: the adapter, thinking levels,
preflight, version probe, sandbox mounts, builtins line, typed tool names,
gate timeout, bundle files, adapter env, turn end and session-files path.
Core reaches a module only through the loader.

A session's route is the instance's harness, model and thinking vars, or
the three fallback. vars; a missing one refuses no-route, and the system
file's execution keys are never read for a session. The launcher checks
the primary route, then the fallback: harness-off, harness-refused,
thinking-unsupported, then preflight. A failed preflight falls back when
a fallback is set, and any refusal of the fallback names the primary's
failure too. Account checks run
only when the route's module counts an account. The manifest records
route, routeReason and harnessVersion, replacing piVersion and
claudeVersion. pi gets --thinking, Claude Code --effort, both from
MOSAIC_THINKING.

The business file refuses a partial fallback. routes and routes.set are
human-only bus verbs outside ORGANIZATION_KEYS; checkRoutes is shared by
routes.set, instances.set and limits.set and checks only changed routes.
The chief of staff changes an instance's model, and the assistant's whole
vars (Q169), through limits.set, so its route write runs the same checks.
A model outside launch.models refuses unknown-model, primary or fallback,
and routes.set passes that code through.
A new run binds only on a switched-on harness module name and records
its transcript path, which the usage reader uses. Settings gains a
Routing table with six selects per instance and loses the Role models
fieldset; it posts changed rows to /api/routes. webui/src/bus.mjs is
unchanged.

Until Jason's `mosaic plugins set` adds "harness": ["claude-code"], a new
launch refuses harness-off.
2026-10-19 11:27:19 -05:00