dewey: row 89 (#1583) PM task reads packet, round 1, manifest b8c5bb3a
Candidate on base a3f6e9e6: tasks takes filters and paging, list_tasks and get_task,
a 40000-character reply cap with compact JSON, and the built-in tools line per harness.
Gate and 34 mutants in the packet.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
This commit is contained in:
Claude Opus 5.5
@@ -0,0 +1,222 @@
|
|||||||
|
# Row 89 (#1583): PM task reads that fit a session, round 1
|
||||||
|
|
||||||
|
Owner: Dewey. Reviewer: Darkwing, then Sage's gate. Brief:
|
||||||
|
`docs/plans/2026-10-12_pm-task-flow.md`, section "PM task reads that fit a
|
||||||
|
session" (blob `e8a1150e`, commit `4d095b77`).
|
||||||
|
|
||||||
|
Base `a3f6e9e6` (queue rev 744). `build.patch` is
|
||||||
|
`git diff --cached --binary a3f6e9e6 --` the twelve paths in
|
||||||
|
`candidate-manifest.sha256`. Apply it with `git apply --index` to a detached
|
||||||
|
worktree of the base; the manifest then checks 12/12 OK. No commit from the
|
||||||
|
base to `refactor` HEAD `70b7b8d5` touches the twelve paths.
|
||||||
|
|
||||||
|
| File | What changed |
|
||||||
|
|---|---|
|
||||||
|
| `packages/bus/src/broker.mjs` | `tasks` takes optional arguments; `#tasks` filters and pages the short form |
|
||||||
|
| `packages/bus/tests/task-page.test.mjs` | new: six tests on a 500-task board |
|
||||||
|
| `packages/bus/README.md` | the `tasks` row and a paragraph on its arguments |
|
||||||
|
| `packages/harness/src/tools.mjs` | `list_tasks` arguments, `get_task`, the `done: false` default, `REPLY_CAP`, `replyText` |
|
||||||
|
| `packages/harness/src/mcp-server.mjs` | replies through `replyText` |
|
||||||
|
| `packages/harness/src/pi-extension.mjs` | replies through `replyText` |
|
||||||
|
| `packages/harness/src/bundle.mjs` | the built-in tools line names the harness's own tools |
|
||||||
|
| `packages/harness/tests/tools.test.mjs` | the schemas, the default, `replyText`'s boundary |
|
||||||
|
| `packages/harness/tests/mcp-server.test.mjs` | compact JSON, the default sent, the cap from the MCP server |
|
||||||
|
| `packages/harness/tests/pi-session.test.mjs` | compact JSON, the cap, `list_tasks` and `get_task` from a real pi session |
|
||||||
|
| `packages/harness/tests/bundle.test.mjs` | the tools line for each harness, and "none" |
|
||||||
|
| `packages/harness/README.md` | the tools line, compact replies, the cap, the two task tools, a limit |
|
||||||
|
|
||||||
|
## The threshold, and the cap
|
||||||
|
|
||||||
|
The bus unit runs `MOSAIC_CLAUDE_BIN=~/.local/bin/claude` (decision 91),
|
||||||
|
which is version 2.1.296 today. I found the threshold in that binary and then
|
||||||
|
confirmed it with a probe.
|
||||||
|
|
||||||
|
- **The binary.** The MCP tool's `maxResultSizeChars` is `1e5`. The function
|
||||||
|
that picks the persist threshold takes the minimum of that and
|
||||||
|
`pH=50000`. A GrowthBook flag, `tengu_velvet_ibis`, can override it per
|
||||||
|
tool. `cachedGrowthBookFeatures` is `{}` in both Claude config files on
|
||||||
|
this host, so no override applies. A result is saved to a file when the
|
||||||
|
total `.length` of its text blocks is over the threshold. The same
|
||||||
|
variable list holds `tjn=200000`, a budget per message across all tool
|
||||||
|
results. `MAX_MCP_OUTPUT_TOKENS` (25000 tokens by default) comes later
|
||||||
|
than 50000 characters for this kind of text.
|
||||||
|
- **The probe** (`probe/`). `probe-server.mjs` is a stdio MCP server whose
|
||||||
|
one tool returns a text block of exactly `n` characters. A headless
|
||||||
|
`claude -p` run of 2.1.296 (haiku, cost $0.0145) called it with 49990,
|
||||||
|
50000 and 50010 (`probe/results.txt`):
|
||||||
|
- 49990 and 50000 came back inline, at full length.
|
||||||
|
- 50010 came back as "Error: result (50,010 characters across 1 line)
|
||||||
|
exceeds maximum allowed tokens. Output has been saved to" a file under
|
||||||
|
`~/.claude/projects/`, outside the workspace. That is the PM's fault.
|
||||||
|
|
||||||
|
So the threshold is "over 50000 characters", counted as JavaScript counts
|
||||||
|
(UTF-16 units). **`REPLY_CAP` is 40000**, 20% under it. `replyText`
|
||||||
|
measures with `.length` too, so a reply of astral characters is counted the
|
||||||
|
way the CLI counts it (the tools test has a surrogate-pair case at 40002).
|
||||||
|
|
||||||
|
Over the cap, the tool throws "reply too large: N characters, over the
|
||||||
|
40000-character cap. Narrow the query: a filter, a smaller limit, or
|
||||||
|
get_task for one task." Both servers report a throw as a tool error
|
||||||
|
(`isError` on MCP, a thrown error on pi). A reply is never cut short.
|
||||||
|
Replies are compact JSON in both servers; that is the cap's input.
|
||||||
|
|
||||||
|
### Sizes on the test board
|
||||||
|
|
||||||
|
500 tasks, titles about 55 characters, descriptions over 2000:
|
||||||
|
|
||||||
|
| Read | Compact | Indented |
|
||||||
|
|---|---|---|
|
||||||
|
| `list_tasks` default (open, 50) | 11,210 | 16,618 |
|
||||||
|
| `limit: 200` | 44,698, refused by the cap | 66,303 |
|
||||||
|
| no arguments, the whole board | 1,518,073 | |
|
||||||
|
| `get_task`, one full row | 3,028 (task 2; 3,055 at most, 365 for a tombstone) | |
|
||||||
|
|
||||||
|
The default page is about a quarter of the cap. `limit: 200` with these
|
||||||
|
titles is over it, so the session gets the cap error and narrows. That's
|
||||||
|
the cap doing its job, and the reason the default is 50.
|
||||||
|
|
||||||
|
## Decisions
|
||||||
|
|
||||||
|
1. **The no-argument read is unchanged.** `tasks` with `{}` returns
|
||||||
|
`#taskRows` as before, so the Console and `mosaic tasks` read what they
|
||||||
|
did. A test compares it with `taskView` for the pm, a human and a
|
||||||
|
reader.
|
||||||
|
2. **The tool always filters.** `ARG_DEFAULTS = { list_tasks: { done: false } }`
|
||||||
|
is merged in `callTool`, under the session's own arguments. A session
|
||||||
|
can't reach the whole board through `list_tasks`, even with no
|
||||||
|
arguments; `done: true` overrides the default.
|
||||||
|
3. **`task_ref` stands alone.** With any other argument it refuses
|
||||||
|
`invalid-request`. A ref not on the board refuses `task-not-found`, a
|
||||||
|
code the broker already has. The row is today's full row, tombstones
|
||||||
|
included.
|
||||||
|
4. **Tombstones are left out of filtered reads.** A tombstone has no
|
||||||
|
title or state to show, and isn't on the board. `get_task` still
|
||||||
|
returns one by ref.
|
||||||
|
5. **Assignees are roles.** The short form lists the roles whose
|
||||||
|
`roles[*].tracker.botId` is among the task's assignees, sorted. A
|
||||||
|
person or another bot isn't a role and doesn't appear. A role whose
|
||||||
|
`tracker` is null holds no task, so `assignee` for it matches nothing.
|
||||||
|
An `assignee` that isn't a role of the business refuses (`toString`
|
||||||
|
included; the check is `Object.hasOwn`).
|
||||||
|
6. **`changed_at` is the row's `at`**, the time the broker last recorded a
|
||||||
|
change, as the Console's board shows it.
|
||||||
|
7. **Paging is in `task_ref` order as text**, the order `#taskRows`
|
||||||
|
already gives (`ORDER BY task_ref`), and `after` compares with `>` on
|
||||||
|
the same strings. `after` needn't be on the board or match the
|
||||||
|
filters; the page starts past it. `total` counts every match, not the
|
||||||
|
page. `next` is the last ref of the page, null when the page ends on
|
||||||
|
the last match.
|
||||||
|
8. **The page sizes live in two files.** `TASK_PAGE` and `TASK_PAGE_MAX`
|
||||||
|
are exported from `broker.mjs` and repeated in `tools.mjs`, so the
|
||||||
|
session's tool server doesn't load the broker. A tools test holds the
|
||||||
|
schema maximum equal to the broker's.
|
||||||
|
9. **The tools line is built in `prompt()`.** For pi it prints the role's
|
||||||
|
tools as before; for Claude Code, `claudeBuiltins({ tools })`, the
|
||||||
|
names the gate gives Claude Code. Its call site is unchanged. With no
|
||||||
|
tools, it says "none" for both.
|
||||||
|
|
||||||
|
## Limits
|
||||||
|
|
||||||
|
- **The cap is per reply.** The CLI's 200000-character budget is per
|
||||||
|
message: six or more replies near the cap in one turn could still be
|
||||||
|
saved to a file. Five at the cap total exactly 200000, which isn't over.
|
||||||
|
- **A full row over 40000 characters can't be read** through `get_task`.
|
||||||
|
The session gets the cap error. The test board's longest full row is
|
||||||
|
3,055; I haven't measured the live board's.
|
||||||
|
- **The threshold belongs to 2.1.296.** `~/.local/bin/claude` updates
|
||||||
|
itself, so a later version can move it. The harness README says so.
|
||||||
|
The probe here rechecks a new version in one run.
|
||||||
|
- **`limit: 200` can exceed the cap** on a real board with long titles,
|
||||||
|
as it does on the test board. The error tells the session to narrow.
|
||||||
|
|
||||||
|
## Gate (my run on the candidate)
|
||||||
|
|
||||||
|
Sequential, in a worktree of `a3f6e9e6` holding the candidate, `TMPDIR` in
|
||||||
|
my scratch, `test-task.sh` with `DOCKER_HOST=unix:///nonexistent`
|
||||||
|
(`gate.sh`, `gate-summary.txt`). The brief's four node suites, plus tasks
|
||||||
|
and control-board, then every `scripts/test-*.sh`.
|
||||||
|
2026-10-12T20:13:35Z to 20:18:12Z, all rc=0:
|
||||||
|
|
||||||
|
| Suite | Result |
|
||||||
|
|---|---|
|
||||||
|
| bus | 108 pass (the six in `task-page.test.mjs` among them) |
|
||||||
|
| harness | 65 pass |
|
||||||
|
| webui | 84 pass |
|
||||||
|
| cli | 115 pass |
|
||||||
|
| tasks | 74 pass |
|
||||||
|
| control-board | 124 pass |
|
||||||
|
| test-auth | 15 passed |
|
||||||
|
| test-conductor | 17 passed |
|
||||||
|
| test-config | 24 passed |
|
||||||
|
| test-discord | 66 passed |
|
||||||
|
| test-extension-package | 18 passed |
|
||||||
|
| test-foundation | 44 passed |
|
||||||
|
| test-queue | 27 passed |
|
||||||
|
| test-release | 14 passed |
|
||||||
|
| test-task | 26 passed; four Docker groups skip (adapter seam, workspace/capability, live task, live recall pair) |
|
||||||
|
|
||||||
|
After the mutants I added one assertion to `packages/harness/tests/tools.test.mjs`
|
||||||
|
(below), so the manifest's hash for that file is newer than this run. No other
|
||||||
|
suite loads that file. The harness suite on the final candidate, 20:43:56Z to
|
||||||
|
20:44:07Z: 65 pass, rc=0 (the last line of `gate-summary.txt`).
|
||||||
|
|
||||||
|
## Mutants
|
||||||
|
|
||||||
|
`mutants.py` swaps one exact string in a second worktree of the base that holds
|
||||||
|
the candidate. It runs the bus or harness node suite, restores the file and
|
||||||
|
reports killed or SURVIVED; an anchor matching anything but once is reported
|
||||||
|
and skipped (none was). 34 mutants, 34 killed in the end. The copy checked
|
||||||
|
12/12 against the manifest afterwards.
|
||||||
|
|
||||||
|
- **Run 1** (`mutants-run1.txt`, from 20:18:19Z). The first 16 were killed.
|
||||||
|
The 17th, `after ignored`, makes `pages()` in `task-page.test.mjs` follow
|
||||||
|
`next` forever. It hung until the harness's 900 s timeout, and the test
|
||||||
|
child it left behind ran on until I stopped it by PID at 20:35Z. My
|
||||||
|
harness's fault: it killed only the runner, not the runner's children.
|
||||||
|
`mutants.py` now starts each run in its own process group, kills the group
|
||||||
|
after 300 s, and reports a hang as a kill.
|
||||||
|
- **Run 2** (`mutants-run2.txt`, 20:35:08Z to 20:43:08Z), the 18 from
|
||||||
|
`after ignored` on. 17 killed, `after ignored` among them as a hang. One
|
||||||
|
survived: `cap counts bytes` (`Buffer.byteLength` for `.length`). The
|
||||||
|
surrogate-pair case at 40002 units is over the cap either way, so it can't
|
||||||
|
tell units from bytes. I added a case that fits by units and not by bytes:
|
||||||
|
19999 emoji, 40000 units and 80002 bytes, fits.
|
||||||
|
- **Run 3** (`mutants-run3.txt`, 20:43:36Z to 20:43:47Z): `cap counts bytes`
|
||||||
|
killed.
|
||||||
|
|
||||||
|
| Mutant | Site | What it undoes |
|
||||||
|
|---|---|---|
|
||||||
|
| no args filtered too | broker | the no-argument read returns the whole projection |
|
||||||
|
| tombstones kept | broker | filtered reads leave tombstones out |
|
||||||
|
| next on an exact last page | broker | `next` is null when the page ends on the last match |
|
||||||
|
| default limit ignored | broker | the page is 50 by default |
|
||||||
|
| limit max unchecked | broker | `limit` over 200 refuses |
|
||||||
|
| assignee by raw id | broker | assignees are role names |
|
||||||
|
| assignees keep persons | broker | a person's assignment shows as no role |
|
||||||
|
| task_ref with other args | broker | `task_ref` stands alone |
|
||||||
|
| missing task_ref is null | broker | a ref not on the board refuses `task-not-found` |
|
||||||
|
| done check dropped | broker | `done` must be a boolean |
|
||||||
|
| done filter dropped | broker | `done` filters |
|
||||||
|
| assignee any string | broker | `assignee` must be a role of the business |
|
||||||
|
| label check dropped | broker | `label` must be a positive integer |
|
||||||
|
| label filter dropped | broker | `label` filters |
|
||||||
|
| requirement pattern dropped | broker | `requirement` must match the id pattern |
|
||||||
|
| requirement filter dropped | broker | `requirement` filters |
|
||||||
|
| after ignored | broker | `after` starts the page (killed by hanging) |
|
||||||
|
| after inclusive | broker | the page starts past `after`, not at it |
|
||||||
|
| total counts the page | broker | `total` counts every match |
|
||||||
|
| description kept | broker | the short form has no description |
|
||||||
|
| tasks keys closed | broker | `tasks` takes its arguments at all |
|
||||||
|
| ARG_DEFAULTS removed | tools | `list_tasks` sends `done: false` |
|
||||||
|
| default overrides the caller | tools | the session's `done` wins over the default |
|
||||||
|
| cap at >= | tools | a reply of exactly 40000 fits |
|
||||||
|
| cap counts bytes | tools | the cap counts UTF-16 units, as the CLI does |
|
||||||
|
| indented JSON | tools | replies are compact |
|
||||||
|
| cap 50000 | tools | the cap is 40000 |
|
||||||
|
| mcp server uncapped | mcp-server | the MCP server applies the cap |
|
||||||
|
| pi uncapped | pi-extension | the pi extension applies the cap |
|
||||||
|
| tool limit max 500 | tools | the schema's `limit` maximum equals the broker's |
|
||||||
|
| get_task not required | tools | `get_task` requires `task_ref` |
|
||||||
|
| get_task without tracker | tools | `get_task` is for tracker holders only |
|
||||||
|
| bundle line pi names | bundle | Claude Code's line names its own tools |
|
||||||
|
| bundle none dropped | bundle | no tools says "none" |
|
||||||
@@ -0,0 +1,774 @@
|
|||||||
|
diff --git a/packages/bus/README.md b/packages/bus/README.md
|
||||||
|
index 69824a30..c5ca68ad 100644
|
||||||
|
--- a/packages/bus/README.md
|
||||||
|
+++ b/packages/bus/README.md
|
||||||
|
@@ -125,7 +125,8 @@ an operator inspect content and delivery evidence after an uncertain result.
|
||||||
|
| `message.read` | `{id}`; recipient holder's acknowledgement; no new holder can acknowledge old holder's delivery. |
|
||||||
|
| `event.emit` | `{kind:'action.allowed',body:{action:'routine',target?},subject?}`; routine observations only. |
|
||||||
|
| `launch.revoke`, `launch.restore` | No args; outside-agent human CLI only. |
|
||||||
|
-| `inbox`, `tasks` | No args; plain JSON projections scoped to capability's business. |
|
||||||
|
+| `inbox` | No args; plain JSON projection scoped to capability's business. |
|
||||||
|
+| `tasks` | Optional args, scoped to capability's business (row 89, #1583). None: every row of the tasks projection (below). `{task_ref}` alone: that one row, or `task-not-found`. Any of `{done,assignee,label,requirement,limit,after}`: `{tasks,total,next}`, the short form (below). An unknown arg or bad value refuses `invalid-request`. |
|
||||||
|
| `agents` | No args; each held role's latest claim row in the capability's business. Human and reader sessions also get `state`, `working` and `inputNeeded` (Agent state, below); a role session gets the claim rows only. |
|
||||||
|
| `trail` | `{subject}`: task ref, decision UUID or message UUID. Linked records in broker write order. |
|
||||||
|
| `decisions` | `{list}`: `roles` or `recent`; human and reader sessions only. Open decisions routed to a role, or the last 50 closed. |
|
||||||
|
@@ -295,7 +296,20 @@ must not pretend a missing/inaccessible task is open or completed. Each row adds
|
||||||
|
computed at each read and never stored. `priority` is the current row's `fields.priority`, null
|
||||||
|
for a tombstone or a read with none (Vikunja's 0 stays 0). `requirement` is the requirement of the
|
||||||
|
first `task.created` for the same Vikunja task id in the business, so a task moved to another
|
||||||
|
-project keeps it; null for a task the stack didn't create. Trail joins
|
||||||
|
+project keeps it; null for a task the stack didn't create.
|
||||||
|
+
|
||||||
|
+A `tasks` read with filters (row 89, #1583) is what a session's `list_tasks` makes; the full
|
||||||
|
+projection is too large for a tool reply on a real board. It leaves out tombstones and pages in
|
||||||
|
+`task_ref` order (as text). Each task is `{task_ref,title,done,bucket,assignees,labels,priority,
|
||||||
|
+requirement,changed_at}`: no description; `assignees` are the roles whose tracker bot is
|
||||||
|
+assigned, sorted, so a person's assignment shows as none; `changed_at` is the row's `at`, as the
|
||||||
|
+Console shows it. Filters: `done` (boolean), `assignee` (a role of the business), `label` (a
|
||||||
|
+label id), `requirement` (`REQ-[A-Z]+-[1-9][0-9]*`). `limit` is 1 to `TASK_PAGE_MAX` (200),
|
||||||
|
+`TASK_PAGE` (50) by default. `after` is a task ref, which needn't be on the board; the page holds
|
||||||
|
+the matches past it. `total` counts every match, `next` is the `after` for the following page,
|
||||||
|
+null on the last.
|
||||||
|
+
|
||||||
|
+Trail joins
|
||||||
|
human request, messages/deliveries, decisions/resolutions, task snapshots and
|
||||||
|
session/claim evidence. Broker timestamps advance monotonically across all
|
||||||
|
its writes, including after reopen; per-table `seq` stays the primary table
|
||||||
|
diff --git a/packages/bus/src/broker.mjs b/packages/bus/src/broker.mjs
|
||||||
|
index 4389b353..4a3e52a4 100644
|
||||||
|
--- a/packages/bus/src/broker.mjs
|
||||||
|
+++ b/packages/bus/src/broker.mjs
|
||||||
|
@@ -42,6 +42,10 @@ const CLOSED = "('resolved','withdrawn','expired')";
|
||||||
|
export const RECENT = 50;
|
||||||
|
// How many replies `replies` returns, oldest first.
|
||||||
|
export const REPLIES = 50;
|
||||||
|
+// Row 89 (#1583): the arguments `tasks` takes, and its page size by default and at most.
|
||||||
|
+const TASK_ARGS = ['done', 'assignee', 'label', 'requirement', 'limit', 'after', 'task_ref'];
|
||||||
|
+export const TASK_PAGE = 50;
|
||||||
|
+export const TASK_PAGE_MAX = 200;
|
||||||
|
export const TASK_VERBS = Object.freeze(ACTIONS.filter((a) => a.startsWith('task.')));
|
||||||
|
const SELF_KINDS = new Set(['task.created', 'task.assigned', 'task.state', 'task.closed', 'task.conflict']);
|
||||||
|
const POLL_KINDS = new Set([
|
||||||
|
@@ -328,6 +332,59 @@ export class Broker {
|
||||||
|
return { ...r, fields, priority, requirement: requirement.get(taskId(r.task_ref)) ?? null };
|
||||||
|
});
|
||||||
|
}
|
||||||
|
+ // Row 89 (#1583): with no arguments, every row as #taskRows gives it, which the Console and
|
||||||
|
+ // `mosaic tasks` read. `task_ref`, alone, gives that one row. Any other argument filters and pages
|
||||||
|
+ // the short form in task_ref order: no description, and the assignees that are a role's tracker
|
||||||
|
+ // bot, by role. A tombstone isn't on the board, so a filtered read leaves it out. `total` counts
|
||||||
|
+ // every task the filters match; `next` is the `after` for the next page, null on the last.
|
||||||
|
+ #tasks(business, a) {
|
||||||
|
+ const rows = this.#taskRows(business);
|
||||||
|
+ if (!Object.keys(a).length) return rows;
|
||||||
|
+ if ('task_ref' in a) {
|
||||||
|
+ if (Object.keys(a).length !== 1 || !taskRef(a.task_ref)) fail('invalid-request');
|
||||||
|
+ return rows.find((r) => r.task_ref === a.task_ref) ?? fail('task-not-found');
|
||||||
|
+ }
|
||||||
|
+ const roles = this.#business(business).roles ?? {};
|
||||||
|
+ const positive = (x, max = Number.MAX_SAFE_INTEGER) => Number.isSafeInteger(x) && x >= 1 && x <= max;
|
||||||
|
+ if (
|
||||||
|
+ ('done' in a && typeof a.done !== 'boolean') ||
|
||||||
|
+ ('assignee' in a && !(typeof a.assignee === 'string' && Object.hasOwn(roles, a.assignee))) ||
|
||||||
|
+ ('label' in a && !positive(a.label)) ||
|
||||||
|
+ ('requirement' in a && !(typeof a.requirement === 'string' && /^REQ-[A-Z]+-[1-9][0-9]*$/.test(a.requirement))) ||
|
||||||
|
+ ('limit' in a && !positive(a.limit, TASK_PAGE_MAX)) ||
|
||||||
|
+ ('after' in a && !taskRef(a.after))
|
||||||
|
+ )
|
||||||
|
+ fail('invalid-request');
|
||||||
|
+ const bots = new Map(
|
||||||
|
+ Object.entries(roles)
|
||||||
|
+ .filter(([, r]) => r.tracker?.botId !== undefined)
|
||||||
|
+ .map(([role, r]) => [r.tracker.botId, role]),
|
||||||
|
+ );
|
||||||
|
+ const matched = rows
|
||||||
|
+ .filter((r) => !r.fields.gone)
|
||||||
|
+ .map(({ task_ref, fields: f, priority, requirement, at }) => ({
|
||||||
|
+ task_ref,
|
||||||
|
+ title: f.title ?? null,
|
||||||
|
+ done: Boolean(f.done),
|
||||||
|
+ bucket: f.bucket ?? null,
|
||||||
|
+ assignees: (f.assignees ?? []).filter((u) => bots.has(u)).map((u) => bots.get(u)).sort(),
|
||||||
|
+ labels: f.labels ?? [],
|
||||||
|
+ priority,
|
||||||
|
+ requirement,
|
||||||
|
+ changed_at: at,
|
||||||
|
+ }))
|
||||||
|
+ .filter(
|
||||||
|
+ (t) =>
|
||||||
|
+ (!('done' in a) || t.done === a.done) &&
|
||||||
|
+ (!('assignee' in a) || t.assignees.includes(a.assignee)) &&
|
||||||
|
+ (!('label' in a) || t.labels.includes(a.label)) &&
|
||||||
|
+ (!('requirement' in a) || t.requirement === a.requirement),
|
||||||
|
+ );
|
||||||
|
+ const rest = 'after' in a ? matched.filter((t) => t.task_ref > a.after) : matched;
|
||||||
|
+ const limit = a.limit ?? TASK_PAGE;
|
||||||
|
+ const tasks = rest.slice(0, limit);
|
||||||
|
+ return { tasks, total: matched.length, next: rest.length > limit ? tasks.at(-1).task_ref : null };
|
||||||
|
+ }
|
||||||
|
#checkAuthority(s, action, { decision = null, target = null } = {}) {
|
||||||
|
this.#agent(s);
|
||||||
|
if (
|
||||||
|
@@ -841,7 +898,8 @@ export class Broker {
|
||||||
|
case 'agents':
|
||||||
|
case 'tasks':
|
||||||
|
case 'trail': {
|
||||||
|
- keys(a, verb === 'trail' ? ['subject'] : [], verb === 'trail' ? ['subject'] : []);
|
||||||
|
+ const allowed = verb === 'trail' ? ['subject'] : verb === 'tasks' ? TASK_ARGS : [];
|
||||||
|
+ keys(a, allowed, verb === 'trail' ? ['subject'] : []);
|
||||||
|
if (!s.human && !s.reader) this.#agent(s);
|
||||||
|
if (verb === 'inbox')
|
||||||
|
return this.#store
|
||||||
|
@@ -859,7 +917,7 @@ export class Broker {
|
||||||
|
// A role session gets the claims only: another role's decisions aren't its to read.
|
||||||
|
return s.human || s.reader ? claims.map((c) => this.#agentState(c)) : claims;
|
||||||
|
}
|
||||||
|
- if (verb === 'tasks') return this.#taskRows(s.business);
|
||||||
|
+ if (verb === 'tasks') return this.#tasks(s.business, a);
|
||||||
|
return this.#trail(s, identifier(a.subject));
|
||||||
|
}
|
||||||
|
// Row 57 (#1546): the Inbox's "Routed to roles" and "Recent" lists, for human and
|
||||||
|
diff --git a/packages/bus/tests/task-page.test.mjs b/packages/bus/tests/task-page.test.mjs
|
||||||
|
new file mode 100644
|
||||||
|
index 00000000..0cfa0598
|
||||||
|
--- /dev/null
|
||||||
|
+++ b/packages/bus/tests/task-page.test.mjs
|
||||||
|
@@ -0,0 +1,225 @@
|
||||||
|
+import test from 'node:test';
|
||||||
|
+import assert from 'node:assert/strict';
|
||||||
|
+import { mkdtempSync, rmSync } from 'node:fs';
|
||||||
|
+import { tmpdir } from 'node:os';
|
||||||
|
+import { join } from 'node:path';
|
||||||
|
+import { Store } from '../src/store.mjs';
|
||||||
|
+import { Broker, TASK_PAGE, TASK_PAGE_MAX } from '../src/broker.mjs';
|
||||||
|
+import { REPLY_CAP } from '../../harness/src/tools.mjs';
|
||||||
|
+// Row 89 (#1583): `tasks` with arguments filters and pages a short form a session can read; with
|
||||||
|
+// none it is the whole board, as the Console and `mosaic tasks` read it.
|
||||||
|
+const authority = { withinRole: ['message.send', 'task.create'], crossRole: [] };
|
||||||
|
+const roles = {
|
||||||
|
+ pm: { authority, tracker: { bot: 'bot-pm', botId: 101 } },
|
||||||
|
+ coder: { authority, tracker: { bot: 'bot-coder', botId: 102 } },
|
||||||
|
+ reviewer: { authority, tracker: null },
|
||||||
|
+};
|
||||||
|
+const businesses = { demo: { id: 'demo', human: 'jason', arbiters: { technical: 'pm', delivery: 'pm' }, roles } };
|
||||||
|
+const at = '2026-10-12T12:00:00.000Z';
|
||||||
|
+const digest = 'd'.repeat(64);
|
||||||
|
+const N = 500;
|
||||||
|
+const words = 'the board holds this task so a session can page through it without the description '.repeat(30);
|
||||||
|
+const ref = (i) => `vikunja:3/${i}`;
|
||||||
|
+// Task i: done when i % 5 == 0; the pm's bot on i % 3 == 0, the coder's on i % 4 == 0, a person (7)
|
||||||
|
+// on i % 2 == 0; label 9 on i % 7 == 0; tombstones at 13 and 26.
|
||||||
|
+const fieldsOf = (i) => ({
|
||||||
|
+ project_id: 3,
|
||||||
|
+ title: `Task ${i}: a title of ordinary length for a tracker card`,
|
||||||
|
+ description: `<p>${i} ${words}</p>`,
|
||||||
|
+ done: i % 5 === 0,
|
||||||
|
+ due_date: null,
|
||||||
|
+ priority: i % 4,
|
||||||
|
+ percent_done: 0,
|
||||||
|
+ bucket: i % 5 === 0 ? 12 : 11,
|
||||||
|
+ labels: i % 7 === 0 ? [9, 4] : [4],
|
||||||
|
+ assignees: [...(i % 3 === 0 ? [101] : []), ...(i % 4 === 0 ? [102] : []), ...(i % 2 === 0 ? [7] : [])],
|
||||||
|
+});
|
||||||
|
+const GONE = new Set([13, 26]);
|
||||||
|
+
|
||||||
|
+function setup(t) {
|
||||||
|
+ const root = mkdtempSync(join(tmpdir(), 'bus-task-page-'));
|
||||||
|
+ const store = new Store(root);
|
||||||
|
+ const b = new Broker({ store, businesses });
|
||||||
|
+ t.after(() => {
|
||||||
|
+ store.close();
|
||||||
|
+ rmSync(root, { recursive: true, force: true });
|
||||||
|
+ });
|
||||||
|
+ const human = b.bindHuman({ business: 'demo', human: 'jason', via: 'cli', outsideAgent: true });
|
||||||
|
+ const pm = b.bindLaunch({ business: 'demo', role: 'pm', run: 'demo-pm', harness: 'claude-code' });
|
||||||
|
+ b.request(pm, { verb: 'role.claim' });
|
||||||
|
+ b.recordTask({
|
||||||
|
+ business: 'demo',
|
||||||
|
+ snapshots: Array.from({ length: N }, (_, k) => k + 1).map((i) => ({
|
||||||
|
+ task_ref: ref(i),
|
||||||
|
+ updated: at,
|
||||||
|
+ digest,
|
||||||
|
+ fields: GONE.has(i) ? { gone: 'no-access' } : fieldsOf(i),
|
||||||
|
+ via: GONE.has(i) ? 'task' : 'board',
|
||||||
|
+ read_at: at,
|
||||||
|
+ })),
|
||||||
|
+ });
|
||||||
|
+ // Task 2 and task 6 were created by the stack, under two requirements.
|
||||||
|
+ for (const [i, requirement] of [[2, 'REQ-SLICE-1'], [6, 'REQ-SLICE-2']]) {
|
||||||
|
+ const { request } = b.request(human, { verb: 'message.send', args: { to: 'pm', body: 'please' } });
|
||||||
|
+ b.recordTask({
|
||||||
|
+ cap: pm,
|
||||||
|
+ business: 'demo',
|
||||||
|
+ snapshots: [{ task_ref: ref(i), updated: at, digest, fields: fieldsOf(i) }],
|
||||||
|
+ events: [{ kind: 'task.created', subject: ref(i), body: { request, requirement } }],
|
||||||
|
+ });
|
||||||
|
+ }
|
||||||
|
+ const tasks = (args, cap = pm) => b.request(cap, { verb: 'tasks', args });
|
||||||
|
+ return { b, human, pm, reader: b.bindReader({ business: 'demo' }), tasks };
|
||||||
|
+}
|
||||||
|
+const refusesWith = (code) => (e) => e.code === code;
|
||||||
|
+// Every task, in the broker's order (task_ref, as text).
|
||||||
|
+const ALL = Array.from({ length: N }, (_, k) => k + 1).sort((x, y) => (ref(x) < ref(y) ? -1 : 1));
|
||||||
|
+const live = (keep) => ALL.filter((i) => !GONE.has(i) && keep(i)).map(ref);
|
||||||
|
+// Every page of a read, by following next.
|
||||||
|
+function pages(tasks, args) {
|
||||||
|
+ const out = [];
|
||||||
|
+ let after;
|
||||||
|
+ for (;;) {
|
||||||
|
+ const page = tasks(after === undefined ? args : { ...args, after });
|
||||||
|
+ out.push(page);
|
||||||
|
+ if (page.next === null) return out;
|
||||||
|
+ after = page.next;
|
||||||
|
+ }
|
||||||
|
+}
|
||||||
|
+
|
||||||
|
+test('the session default fits under the reply cap and pages to the end with next', (t) => {
|
||||||
|
+ const { tasks } = setup(t);
|
||||||
|
+ const whole = tasks({});
|
||||||
|
+ assert.equal(whole.length, N);
|
||||||
|
+ assert.ok(JSON.stringify(whole).length > 20 * REPLY_CAP, 'the whole board is far over the cap');
|
||||||
|
+ const read = pages(tasks, { done: false });
|
||||||
|
+ const open = live((i) => i % 5 !== 0);
|
||||||
|
+ assert.equal(read.length, Math.ceil(open.length / TASK_PAGE));
|
||||||
|
+ for (const page of read) {
|
||||||
|
+ assert.ok(JSON.stringify(page).length < REPLY_CAP, `${JSON.stringify(page).length}`);
|
||||||
|
+ assert.equal(page.total, open.length);
|
||||||
|
+ assert.ok(page.tasks.length <= TASK_PAGE);
|
||||||
|
+ }
|
||||||
|
+ assert.deepEqual(read.flatMap((p) => p.tasks.map((x) => x.task_ref)), open);
|
||||||
|
+ assert.deepEqual(read.slice(0, -1).map((p) => p.next), read.slice(0, -1).map((p) => p.tasks.at(-1).task_ref));
|
||||||
|
+ assert.equal(read.at(-1).next, null);
|
||||||
|
+ // limit at its most pages the same tasks in fewer pages.
|
||||||
|
+ const big = pages(tasks, { done: false, limit: TASK_PAGE_MAX });
|
||||||
|
+ assert.equal(big.length, Math.ceil(open.length / TASK_PAGE_MAX));
|
||||||
|
+ assert.deepEqual(big.flatMap((p) => p.tasks.map((x) => x.task_ref)), open);
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
+test('the short form: no description, assignees by role, every listed key', (t) => {
|
||||||
|
+ const { tasks } = setup(t);
|
||||||
|
+ const whole = new Map(tasks({}).map((r) => [r.task_ref, r]));
|
||||||
|
+ const page = tasks({ limit: TASK_PAGE_MAX });
|
||||||
|
+ for (const x of page.tasks) {
|
||||||
|
+ const row = whole.get(x.task_ref);
|
||||||
|
+ const i = Number(x.task_ref.split('/')[1]);
|
||||||
|
+ assert.deepEqual(x, {
|
||||||
|
+ task_ref: row.task_ref,
|
||||||
|
+ title: row.fields.title,
|
||||||
|
+ done: row.fields.done,
|
||||||
|
+ bucket: row.fields.bucket,
|
||||||
|
+ assignees: [...(i % 4 === 0 ? ['coder'] : []), ...(i % 3 === 0 ? ['pm'] : [])],
|
||||||
|
+ labels: row.fields.labels,
|
||||||
|
+ priority: row.priority,
|
||||||
|
+ requirement: row.requirement,
|
||||||
|
+ changed_at: row.at,
|
||||||
|
+ });
|
||||||
|
+ }
|
||||||
|
+ const two = page.tasks.find((x) => x.task_ref === ref(2));
|
||||||
|
+ assert.equal(two.requirement, 'REQ-SLICE-1');
|
||||||
|
+ assert.equal(JSON.stringify(page).includes('description'), false);
|
||||||
|
+ // A tombstone is on the whole board, never in a filtered read.
|
||||||
|
+ assert.deepEqual(whole.get(ref(13)).fields, { gone: 'no-access' });
|
||||||
|
+ const every = pages(tasks, { limit: TASK_PAGE_MAX }).flatMap((p) => p.tasks.map((x) => x.task_ref));
|
||||||
|
+ assert.deepEqual(every, live(() => true));
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
+test('each filter, alone and together, and total counts every match', (t) => {
|
||||||
|
+ const { tasks } = setup(t);
|
||||||
|
+ const refs = (args) => pages(tasks, { ...args, limit: TASK_PAGE_MAX }).flatMap((p) => p.tasks.map((x) => x.task_ref));
|
||||||
|
+ assert.deepEqual(refs({ done: true }), live((i) => i % 5 === 0));
|
||||||
|
+ assert.deepEqual(refs({ done: false }), live((i) => i % 5 !== 0));
|
||||||
|
+ assert.deepEqual(refs({ assignee: 'pm' }), live((i) => i % 3 === 0));
|
||||||
|
+ assert.deepEqual(refs({ assignee: 'coder' }), live((i) => i % 4 === 0));
|
||||||
|
+ // A role with no tracker bot holds no task; a person's assignment is no role's.
|
||||||
|
+ assert.deepEqual(refs({ assignee: 'reviewer' }), []);
|
||||||
|
+ assert.deepEqual(refs({ label: 9 }), live((i) => i % 7 === 0));
|
||||||
|
+ assert.deepEqual(refs({ label: 4 }), live(() => true));
|
||||||
|
+ assert.deepEqual(refs({ label: 5 }), []);
|
||||||
|
+ assert.deepEqual(refs({ requirement: 'REQ-SLICE-2' }), [ref(6)]);
|
||||||
|
+ assert.deepEqual(refs({ requirement: 'REQ-NONE-1' }), []);
|
||||||
|
+ assert.deepEqual(refs({ done: false, assignee: 'pm', label: 9 }), live((i) => i % 5 !== 0 && i % 3 === 0 && i % 7 === 0));
|
||||||
|
+ const page = tasks({ done: true, assignee: 'coder', limit: 3 });
|
||||||
|
+ assert.equal(page.total, live((i) => i % 5 === 0 && i % 4 === 0).length);
|
||||||
|
+ assert.equal(page.tasks.length, 3);
|
||||||
|
+ assert.equal(page.next, page.tasks[2].task_ref);
|
||||||
|
+ // after needn't be a task on the board, or one the filters match: the page starts past it, in
|
||||||
|
+ // text order (3/500 < 3/501 < 3/51), and total still counts every match.
|
||||||
|
+ const past = tasks({ after: 'vikunja:3/501', limit: 2 });
|
||||||
|
+ assert.deepEqual(past.tasks.map((x) => x.task_ref), [ref(51), ref(52)]);
|
||||||
|
+ assert.equal(past.total, live(() => true).length);
|
||||||
|
+ assert.deepEqual(tasks({ done: true, after: ref(51), limit: 1 }).tasks.map((x) => x.task_ref), [ref(55)]);
|
||||||
|
+ const last = tasks({ after: 'vikunja:9/1' });
|
||||||
|
+ assert.deepEqual(last, { tasks: [], total: live(() => true).length, next: null });
|
||||||
|
+ // A page that ends exactly on the last match has no next.
|
||||||
|
+ const exact = tasks({ requirement: 'REQ-SLICE-1', limit: 1 });
|
||||||
|
+ assert.deepEqual([exact.tasks.length, exact.next], [1, null]);
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
+test('task_ref reads one task in full, as the whole board has it', (t) => {
|
||||||
|
+ const { tasks, human, reader } = setup(t);
|
||||||
|
+ const whole = tasks({});
|
||||||
|
+ for (const r of [ref(2), ref(13), ref(500)]) assert.deepEqual(tasks({ task_ref: r }), whole.find((x) => x.task_ref === r));
|
||||||
|
+ assert.ok(tasks({ task_ref: ref(2) }).fields.description.length > 2000);
|
||||||
|
+ assert.deepEqual(tasks({ task_ref: ref(2) }, human), tasks({ task_ref: ref(2) }, reader));
|
||||||
|
+ assert.throws(() => tasks({ task_ref: 'vikunja:3/501' }), refusesWith('task-not-found'));
|
||||||
|
+ assert.throws(() => tasks({ task_ref: 'vikunja:4/2' }), refusesWith('task-not-found'));
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
+test('no arguments is the read the Console and mosaic make, unchanged', (t) => {
|
||||||
|
+ const { b, tasks, human, reader } = setup(t);
|
||||||
|
+ const rows = b.taskView('demo', 'current');
|
||||||
|
+ const whole = tasks({});
|
||||||
|
+ assert.deepEqual(whole.map(({ priority, requirement, ...r }) => r), rows);
|
||||||
|
+ assert.deepEqual(tasks({}, human), whole);
|
||||||
|
+ assert.deepEqual(tasks({}, reader), whole);
|
||||||
|
+ assert.deepEqual(b.request(reader, { verb: 'tasks' }), whole);
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
+test('an unknown argument or a bad value refuses invalid-request', (t) => {
|
||||||
|
+ const { tasks, reader } = setup(t);
|
||||||
|
+ const bad = [
|
||||||
|
+ { status: 'open' },
|
||||||
|
+ { description: 'x' },
|
||||||
|
+ { done: 'false' },
|
||||||
|
+ { done: 0 },
|
||||||
|
+ { done: null },
|
||||||
|
+ { assignee: 'cto' },
|
||||||
|
+ { assignee: 'toString' },
|
||||||
|
+ { assignee: 101 },
|
||||||
|
+ { label: 0 },
|
||||||
|
+ { label: -1 },
|
||||||
|
+ { label: 1.5 },
|
||||||
|
+ { label: '9' },
|
||||||
|
+ { requirement: 'REQ-slice-1' },
|
||||||
|
+ { requirement: 'REQ-SLICE-0' },
|
||||||
|
+ { requirement: 7 },
|
||||||
|
+ { limit: 0 },
|
||||||
|
+ { limit: TASK_PAGE_MAX + 1 },
|
||||||
|
+ { limit: 2.5 },
|
||||||
|
+ { limit: '50' },
|
||||||
|
+ { after: 'vikunja:3' },
|
||||||
|
+ { after: 3 },
|
||||||
|
+ { task_ref: 'vikunja:3/x' },
|
||||||
|
+ { task_ref: ref(2), done: false },
|
||||||
|
+ { task_ref: ref(2), limit: 1 },
|
||||||
|
+ ];
|
||||||
|
+ for (const args of bad) {
|
||||||
|
+ assert.throws(() => tasks(args), refusesWith('invalid-request'), JSON.stringify(args));
|
||||||
|
+ assert.throws(() => tasks(args, reader), refusesWith('invalid-request'), JSON.stringify(args));
|
||||||
|
+ }
|
||||||
|
+ assert.deepEqual([TASK_PAGE, TASK_PAGE_MAX], [50, 200]);
|
||||||
|
+});
|
||||||
|
diff --git a/packages/harness/README.md b/packages/harness/README.md
|
||||||
|
index 7598a573..1d1d7c5b 100644
|
||||||
|
--- a/packages/harness/README.md
|
||||||
|
+++ b/packages/harness/README.md
|
||||||
|
@@ -41,7 +41,10 @@ writes the bundle once into `dir` (0700; every file 0600, created with `wx`,
|
||||||
|
so an existing bundle is never overwritten):
|
||||||
|
|
||||||
|
- `prompt.md`: the role contract, then a generated "This session" section
|
||||||
|
- (business, instance, run, who launched it, workspace, typed tools).
|
||||||
|
+ (business, instance, run, who launched it, workspace, built-in and typed
|
||||||
|
+ tools). The built-in tools are named as the harness names them, as the
|
||||||
|
+ manifest lists them: `limits.tools` for Pi, `claudeBuiltins` for Claude
|
||||||
|
+ Code (Read, Write, Edit, Grep, Glob).
|
||||||
|
- `policy.json`: harness, workspace, built-in tools (`limits.tools`), typed
|
||||||
|
tool names, network, the services the role needs, and the credential
|
||||||
|
status the host read from the broker (metadata only, never a token).
|
||||||
|
@@ -71,6 +74,20 @@ The harness never holds the capability. Tools call the runner's tool socket
|
||||||
|
against `tools.json` and forwards the call with its own capability. A
|
||||||
|
refusal comes back to the model as `refused: <code>`.
|
||||||
|
|
||||||
|
+A reply goes to the model as compact JSON, from the MCP server and the Pi
|
||||||
|
+extension alike (`replyText` in `tools.mjs`). Over `REPLY_CAP`, 40000
|
||||||
|
+characters, the tool returns an error that names the reply's size and the
|
||||||
|
+cap and asks the session to narrow the query, never a cut-off reply. The
|
||||||
|
+pinned Claude CLI (2.1.296) saves an MCP result over 50000 characters to a
|
||||||
|
+file outside the workspace, which the gate refuses to read (row 89, #1583;
|
||||||
|
+`agents/dewey/work/queue-89/build.md` has how that size was found).
|
||||||
|
+
|
||||||
|
+`list_tasks` (tracker holders) takes the broker's `tasks` filters, `done`,
|
||||||
|
+`assignee`, `label`, `requirement`, `limit` and `after`, and sends
|
||||||
|
+`done: false` when the session leaves `done` out, so its default is open
|
||||||
|
+tasks, 50 at a time, without descriptions. `get_task` takes a `task_ref`
|
||||||
|
+and returns that task in full. `packages/bus/README.md` has the reply.
|
||||||
|
+
|
||||||
|
## The gate
|
||||||
|
|
||||||
|
`gate.mjs` `decide(policy, tool, input)` answers one tool call: the
|
||||||
|
@@ -188,6 +205,13 @@ These are limits, not bugs, and nothing in this package claims otherwise.
|
||||||
|
- **Resolution** runs without the project layer, there is no skills source
|
||||||
|
(bundles list none), and the resolved `thinking` level is recorded in the
|
||||||
|
manifest but not applied to either harness.
|
||||||
|
+- **The reply cap is per reply.** Claude Code also persists the largest
|
||||||
|
+ results of one message when their total passes 200000 characters, so
|
||||||
|
+ six or more replies near the cap in one turn can still be saved to a
|
||||||
|
+ file. A single task whose full row is over the cap (a description near
|
||||||
|
+ 40000 characters) can't be read through `get_task`. The cap matches
|
||||||
|
+ Claude CLI 2.1.296; the binary the bus unit names updates itself, so a
|
||||||
|
+ later version can move the threshold.
|
||||||
|
- **`--restricted`** (Claude Code) is not one of the S0 lines. It is what
|
||||||
|
keeps `CLAUDE.md` files and auto-memory out of the prompt (above).
|
||||||
|
- **Other spellings.** A read is refused when a spelling Pi might open
|
||||||
|
diff --git a/packages/harness/src/bundle.mjs b/packages/harness/src/bundle.mjs
|
||||||
|
index 2f16b1c8..9326be87 100644
|
||||||
|
--- a/packages/harness/src/bundle.mjs
|
||||||
|
+++ b/packages/harness/src/bundle.mjs
|
||||||
|
@@ -64,7 +64,8 @@ function prompt({ contract, resolved, session, tools, untyped }) {
|
||||||
|
`- Business: ${resolved.business}. Role instance: ${resolved.instance} (definition ${resolved.definition}). Run: ${session.run}.`,
|
||||||
|
`- Launched by: ${session.launchedBy}. Harness: ${session.harness}, model ${session.model}.`,
|
||||||
|
`- Workspace: ${session.workspace}. File tools work inside it only.`,
|
||||||
|
- `- Built-in tools: ${resolved.limits.tools.length ? resolved.limits.tools.join(", ") : "none"}.`,
|
||||||
|
+ // Row 89 (#1583): the names the harness gives its built-ins, as the manifest lists them.
|
||||||
|
+ `- Built-in tools: ${(session.harness === "pi" ? resolved.limits.tools : claudeBuiltins({ tools: resolved.limits.tools })).join(", ") || "none"}.`,
|
||||||
|
`- Bus tools: ${tools.map((t) => t.name).join(", ")}. The broker decides every call; a refusal names its reason.`,
|
||||||
|
"- Each message to you arrives as one turn. Your final answer goes back to the sender as the reply, so end each turn with the answer itself.",
|
||||||
|
"- You never hold a token. You act on the tracker, the forge and other roles only through the bus tools.",
|
||||||
|
diff --git a/packages/harness/src/mcp-server.mjs b/packages/harness/src/mcp-server.mjs
|
||||||
|
index 0f43948e..5415a55e 100644
|
||||||
|
--- a/packages/harness/src/mcp-server.mjs
|
||||||
|
+++ b/packages/harness/src/mcp-server.mjs
|
||||||
|
@@ -1,11 +1,12 @@
|
||||||
|
// Minimal stdio MCP server that gives a Claude Code session its typed
|
||||||
|
// tools: `node mcp-server.mjs <tools.json> <tool socket>`. Newline-delimited
|
||||||
|
// JSON-RPC 2.0 with initialize, tools/list, tools/call and ping. Each call
|
||||||
|
-// goes to the runner's tool socket; the server holds no capability.
|
||||||
|
+// goes to the runner's tool socket; the server holds no capability. A
|
||||||
|
+// reply over REPLY_CAP is a tool error (tools.mjs, replyText).
|
||||||
|
|
||||||
|
import { readFileSync } from "node:fs";
|
||||||
|
import { createInterface } from "node:readline";
|
||||||
|
-import { callTool } from "./tools.mjs";
|
||||||
|
+import { callTool, replyText } from "./tools.mjs";
|
||||||
|
|
||||||
|
const [toolsFile, socket] = process.argv.slice(2);
|
||||||
|
if (!toolsFile || !socket) {
|
||||||
|
@@ -37,7 +38,7 @@ async function handle(m) {
|
||||||
|
if (!byName.has(name)) return error(id, -32602, `unknown tool: ${name}`);
|
||||||
|
try {
|
||||||
|
const result = await callTool(socket, name, m.params?.arguments ?? {});
|
||||||
|
- return reply(id, { content: [{ type: "text", text: JSON.stringify(result, null, 2) }] });
|
||||||
|
+ return reply(id, { content: [{ type: "text", text: replyText(result) }] });
|
||||||
|
} catch (e) {
|
||||||
|
return reply(id, { content: [{ type: "text", text: e.message }], isError: true });
|
||||||
|
}
|
||||||
|
diff --git a/packages/harness/src/pi-extension.mjs b/packages/harness/src/pi-extension.mjs
|
||||||
|
index 2b3b0044..5647f0f5 100644
|
||||||
|
--- a/packages/harness/src/pi-extension.mjs
|
||||||
|
+++ b/packages/harness/src/pi-extension.mjs
|
||||||
|
@@ -10,7 +10,7 @@
|
||||||
|
|
||||||
|
import { readFileSync, writeFileSync } from "node:fs";
|
||||||
|
import { decide } from "./gate.mjs";
|
||||||
|
-import { callTool } from "./tools.mjs";
|
||||||
|
+import { callTool, replyText } from "./tools.mjs";
|
||||||
|
|
||||||
|
function need(name) {
|
||||||
|
const value = process.env[name];
|
||||||
|
@@ -40,9 +40,9 @@ export default function (pi) {
|
||||||
|
description: t.description,
|
||||||
|
parameters: t.parameters,
|
||||||
|
async execute(_toolCallId, params) {
|
||||||
|
- // A refusal throws, which pi reports to the model as a tool error.
|
||||||
|
+ // A refusal, or a reply over REPLY_CAP, throws, which pi reports to the model as a tool error.
|
||||||
|
const result = await callTool(socket, t.name, params ?? {});
|
||||||
|
- return { content: [{ type: "text", text: JSON.stringify(result, null, 2) }], details: {} };
|
||||||
|
+ return { content: [{ type: "text", text: replyText(result) }], details: {} };
|
||||||
|
},
|
||||||
|
});
|
||||||
|
}
|
||||||
|
diff --git a/packages/harness/src/tools.mjs b/packages/harness/src/tools.mjs
|
||||||
|
index 903f3bef..f0011d30 100644
|
||||||
|
--- a/packages/harness/src/tools.mjs
|
||||||
|
+++ b/packages/harness/src/tools.mjs
|
||||||
|
@@ -16,6 +16,11 @@ const str = (description, extra = {}) => ({ type: "string", description, ...extr
|
||||||
|
const obj = (properties, required = []) => ({ type: "object", properties, required, additionalProperties: false });
|
||||||
|
|
||||||
|
const TASK_REF = str("Task reference, as list_tasks shows it");
|
||||||
|
+// The broker's page size for `tasks`, by default and at most (TASK_PAGE and TASK_PAGE_MAX in
|
||||||
|
+// packages/bus/src/broker.mjs; a test holds them equal). Kept here so the session's tool server
|
||||||
|
+// doesn't load the broker.
|
||||||
|
+const TASK_PAGE = 50;
|
||||||
|
+const TASK_PAGE_MAX = 200;
|
||||||
|
const COMMON = {
|
||||||
|
task_ref: TASK_REF,
|
||||||
|
decision: str("Id of a resolved decision that approves this call; needed when the action is cross-role or gated for you"),
|
||||||
|
@@ -113,10 +118,39 @@ export function typedTools({ resolved, arbiter = false, tracker = false }) {
|
||||||
|
add("list_agents", "agents", null, "List the role instances that hold a claim now.", obj({}));
|
||||||
|
add("inbox", "inbox", null, "List the open decisions routed to you.", obj({}));
|
||||||
|
add("trail", "trail", null, "Show the evidence trail for a subject: a run, message, decision or task.", obj({ subject: str("Subject id") }, ["subject"]));
|
||||||
|
- if (tracker) add("list_tasks", "tasks", null, "List the business's tracker tasks as the broker last saw them.", obj({}));
|
||||||
|
+ if (tracker) {
|
||||||
|
+ add("list_tasks", "tasks", null, `List the business's tracker tasks as the broker last saw them, in short form with no description. By default open tasks (done false), ${TASK_PAGE} at a time; pass the reply's next as after for the next page. get_task reads one task in full.`, obj({
|
||||||
|
+ done: { type: "boolean", description: "true for closed tasks, false (the default) for open ones" },
|
||||||
|
+ assignee: str("Only tasks assigned to this role"),
|
||||||
|
+ label: { type: "integer", minimum: 1, description: "Only tasks with this label id" },
|
||||||
|
+ requirement: str("Only tasks created for this requirement id", { pattern: "^REQ-[A-Z]+-[1-9][0-9]*$" }),
|
||||||
|
+ limit: { type: "integer", minimum: 1, maximum: TASK_PAGE_MAX, description: `Tasks per page, ${TASK_PAGE} by default` },
|
||||||
|
+ after: str("The next value of the previous page"),
|
||||||
|
+ }));
|
||||||
|
+ add("get_task", "tasks", null, "Read one tracker task in full, description included.", obj({ task_ref: TASK_REF }, ["task_ref"]));
|
||||||
|
+ }
|
||||||
|
return tools;
|
||||||
|
}
|
||||||
|
|
||||||
|
+// Arguments a tool sends when the session leaves them out. list_tasks always filters: the
|
||||||
|
+// broker's `tasks` with no arguments is the whole board, which is the Console's read.
|
||||||
|
+const ARG_DEFAULTS = { list_tasks: { done: false } };
|
||||||
|
+
|
||||||
|
+// The most characters a tool reply may hold. The pinned Claude CLI (2.1.296, decision 91) saves
|
||||||
|
+// an MCP result over 50000 characters to a file outside the workspace, which the run's file guard
|
||||||
|
+// refuses to read (#1583; agents/dewey/work/queue-89/build.md has how that was found). The cap keeps
|
||||||
|
+// 20% under it, counts as JavaScript does (UTF-16 units, as the CLI does), and holds for pi too.
|
||||||
|
+export const REPLY_CAP = 40_000;
|
||||||
|
+
|
||||||
|
+// The text a tool returns to the model: compact JSON, or an error over the cap. A reply is never
|
||||||
|
+// cut short.
|
||||||
|
+export function replyText(result) {
|
||||||
|
+ const text = JSON.stringify(result) ?? "null";
|
||||||
|
+ if (text.length > REPLY_CAP)
|
||||||
|
+ throw new Error(`reply too large: ${text.length} characters, over the ${REPLY_CAP}-character cap. Narrow the query: a filter, a smaller limit, or get_task for one task.`);
|
||||||
|
+ return text;
|
||||||
|
+}
|
||||||
|
+
|
||||||
|
// Held actions that have no typed tool. The manifest lists them so a
|
||||||
|
// reviewer sees what the session can't do through the bus.
|
||||||
|
export function untypedActions(resolved, tools) {
|
||||||
|
@@ -137,7 +171,7 @@ export function callTool(path, tool, args = {}, { timeout = 60_000 } = {}) {
|
||||||
|
error ? reject(error) : resolve(value);
|
||||||
|
};
|
||||||
|
socket.setTimeout(timeout, () => end(new Error("tool call timed out")));
|
||||||
|
- socket.on("connect", () => socket.write(JSON.stringify({ tool, args }) + "\n"));
|
||||||
|
+ socket.on("connect", () => socket.write(JSON.stringify({ tool, args: { ...ARG_DEFAULTS[tool], ...args } }) + "\n"));
|
||||||
|
socket.on("error", (e) => end(new Error(`tool socket: ${e.code ?? e.message}`)));
|
||||||
|
socket.on("end", () => end(new Error("tool socket closed without a reply")));
|
||||||
|
socket.on("data", (b) => {
|
||||||
|
diff --git a/packages/harness/tests/bundle.test.mjs b/packages/harness/tests/bundle.test.mjs
|
||||||
|
index 23deabf2..98cee8ed 100644
|
||||||
|
--- a/packages/harness/tests/bundle.test.mjs
|
||||||
|
+++ b/packages/harness/tests/bundle.test.mjs
|
||||||
|
@@ -87,6 +87,20 @@ test("a claude-code bundle adds the wrapped gate hook and the MCP config", (t) =
|
||||||
|
assert.deepEqual(Object.keys(m.sources).sort(), ["claudeGate", "gate", "mcpServer", "tools"]);
|
||||||
|
});
|
||||||
|
|
||||||
|
+// Row 89 (#1583): the prompt names the built-ins as the session's harness names them.
|
||||||
|
+test("the prompt's built-in tools line is the harness's own names, as the manifest lists them", (t) => {
|
||||||
|
+ const line = (b) => readFileSync(b.files.prompt, "utf8").split("\n").find((l) => l.startsWith("- Built-in tools: "));
|
||||||
|
+ const pi = build(t, resolvedFor(t, "pm"));
|
||||||
|
+ assert.equal(line(pi), "- Built-in tools: read, write, edit, grep, find, ls.");
|
||||||
|
+ const claude = build(t, resolvedFor(t, "pm", (d) => (d.roles.pm.vars = { ...d.roles.pm.vars, harness: "claude-code", model: "claude-opus-5-5" })));
|
||||||
|
+ assert.equal(line(claude), "- Built-in tools: Read, Write, Edit, Grep, Glob.");
|
||||||
|
+ for (const b of [pi, claude]) assert.equal(line(b), `- Built-in tools: ${b.manifest.builtins.join(", ")}.`);
|
||||||
|
+ for (const harness of ["pi", "claude-code"]) {
|
||||||
|
+ const none = build(t, resolvedFor(t, "pm", (d) => (d.roles.pm.vars = { ...d.roles.pm.vars, harness, model: "claude-opus-5-5", "limits.tools": [] })));
|
||||||
|
+ assert.equal(line(none), "- Built-in tools: none.", harness);
|
||||||
|
+ }
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
test("a bundle is written once: an existing file refuses", (t) => {
|
||||||
|
const dir = scratch(t);
|
||||||
|
mkdirSync(join(dir, "bundle"));
|
||||||
|
diff --git a/packages/harness/tests/mcp-server.test.mjs b/packages/harness/tests/mcp-server.test.mjs
|
||||||
|
index 72992dde..2219730c 100644
|
||||||
|
--- a/packages/harness/tests/mcp-server.test.mjs
|
||||||
|
+++ b/packages/harness/tests/mcp-server.test.mjs
|
||||||
|
@@ -5,7 +5,7 @@ import { writeFileSync } from "node:fs";
|
||||||
|
import { join } from "node:path";
|
||||||
|
import { createInterface } from "node:readline";
|
||||||
|
import { SOURCES } from "../src/bundle.mjs";
|
||||||
|
-import { typedTools } from "../src/tools.mjs";
|
||||||
|
+import { REPLY_CAP, typedTools } from "../src/tools.mjs";
|
||||||
|
import { fakeToolSocket, resolvedFor, scratch } from "./helpers.mjs";
|
||||||
|
|
||||||
|
// The server over stdio: send(message) resolves with the reply to its id.
|
||||||
|
@@ -43,6 +43,8 @@ async function setup(t) {
|
||||||
|
writeFileSync(toolsFile, JSON.stringify(tools));
|
||||||
|
const sock = await fakeToolSocket(t, dir, (tool, args) => {
|
||||||
|
if (tool === "launch") throw new Error("capacity-full");
|
||||||
|
+ // trail answers a string `size` characters long once JSON-encoded, for the reply cap.
|
||||||
|
+ if (tool === "trail" && /^size:/.test(args.subject)) return "x".repeat(Number(args.subject.slice(5)) - 2);
|
||||||
|
return { tool, args };
|
||||||
|
});
|
||||||
|
return { dir, tools, sock, server: serve(t, toolsFile, sock.path) };
|
||||||
|
@@ -70,6 +72,25 @@ test("tools/call goes through the tool socket; a refusal is an isError result",
|
||||||
|
assert.deepEqual(sock.calls.map((c) => c.tool), ["list_agents", "launch"]);
|
||||||
|
});
|
||||||
|
|
||||||
|
+test("replies are compact JSON; over the cap a tool error names the size and the cap", async (t) => {
|
||||||
|
+ const { sock, server } = await setup(t);
|
||||||
|
+ const call = async (name, args) => (await server.send("tools/call", { name, arguments: args })).result;
|
||||||
|
+ const ok = await call("get_task", { task_ref: "vikunja:3/7" });
|
||||||
|
+ assert.equal(ok.content[0].text, '{"tool":"get_task","args":{"task_ref":"vikunja:3/7"}}');
|
||||||
|
+ // list_tasks with no arguments asks the broker for open tasks.
|
||||||
|
+ assert.equal((await call("list_tasks", {})).content[0].text, '{"tool":"list_tasks","args":{"done":false}}');
|
||||||
|
+ const fits = await call("trail", { subject: `size:${REPLY_CAP}` });
|
||||||
|
+ assert.equal(fits.isError, undefined);
|
||||||
|
+ assert.equal(fits.content[0].text.length, REPLY_CAP);
|
||||||
|
+ const over = await call("trail", { subject: `size:${REPLY_CAP + 1}` });
|
||||||
|
+ assert.equal(over.isError, true);
|
||||||
|
+ assert.deepEqual(over.content, [{
|
||||||
|
+ type: "text",
|
||||||
|
+ text: `reply too large: ${REPLY_CAP + 1} characters, over the ${REPLY_CAP}-character cap. Narrow the query: a filter, a smaller limit, or get_task for one task.`,
|
||||||
|
+ }]);
|
||||||
|
+ assert.deepEqual(sock.calls.map((c) => c.tool), ["get_task", "list_tasks", "trail", "trail"]);
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
test("unknown tools and methods are JSON-RPC errors and never reach the socket", async (t) => {
|
||||||
|
const { sock, server } = await setup(t);
|
||||||
|
const unknown = await server.send("tools/call", { name: "resolve_decision", arguments: {} });
|
||||||
|
diff --git a/packages/harness/tests/pi-session.test.mjs b/packages/harness/tests/pi-session.test.mjs
|
||||||
|
index 32db2aa3..95954b9d 100644
|
||||||
|
--- a/packages/harness/tests/pi-session.test.mjs
|
||||||
|
+++ b/packages/harness/tests/pi-session.test.mjs
|
||||||
|
@@ -8,6 +8,7 @@ import { existsSync, mkdirSync, readdirSync, symlinkSync, writeFileSync } from "
|
||||||
|
import { join } from "node:path";
|
||||||
|
import { buildBundle } from "../src/bundle.mjs";
|
||||||
|
import { adapterEnv } from "../src/runner.mjs";
|
||||||
|
+import { REPLY_CAP } from "../src/tools.mjs";
|
||||||
|
import { fakeToolSocket, mockAnthropic, REPO, resolvedFor, scratch } from "./helpers.mjs";
|
||||||
|
|
||||||
|
const ADAPTER = join(REPO, "adapters", "pi", "adapter.sh");
|
||||||
|
@@ -25,6 +26,8 @@ async function session(t, script, place = (dir) => join(dir, "ws")) {
|
||||||
|
const resolved = resolvedFor(t, "pm", (d) => (d.roles.pm.vars = { ...d.roles.pm.vars, model: "probe-model" }));
|
||||||
|
const sock = await fakeToolSocket(t, dir, (tool, args) => {
|
||||||
|
if (tool === "launch") throw new Error("capacity-full");
|
||||||
|
+ // trail answers a string `size` characters long once JSON-encoded, for the reply cap.
|
||||||
|
+ if (tool === "trail" && /^size:/.test(args.subject)) return "x".repeat(Number(args.subject.slice(5)) - 2);
|
||||||
|
return { tool, args };
|
||||||
|
});
|
||||||
|
const s = {
|
||||||
|
@@ -78,7 +81,7 @@ test("pi: typed tools reach the socket, the gate blocks, agent_end writes the ma
|
||||||
|
assert.equal(results.length, 5);
|
||||||
|
// list_agents went through the socket; launch was refused by it.
|
||||||
|
assert.equal(results[0][0], false);
|
||||||
|
- assert.match(results[0][1], /"tool": "list_agents"/);
|
||||||
|
+ assert.match(results[0][1], /^\{"tool":"list_agents"/);
|
||||||
|
assert.equal(results[1][0], true);
|
||||||
|
assert.match(results[1][1], /refused: capacity-full/);
|
||||||
|
// read inside passes; outside is gated; bash isn't in the PM's tools.
|
||||||
|
@@ -96,6 +99,30 @@ test("pi: typed tools reach the socket, the gate blocks, agent_end writes the ma
|
||||||
|
assert.ok(readdirSync(s.sessionDir).length > 0);
|
||||||
|
});
|
||||||
|
|
||||||
|
+test("pi: a reply over the cap is a tool error naming the size and the cap; list_tasks asks for open tasks", async (t) => {
|
||||||
|
+ const { sock, s, env } = await session(t, [
|
||||||
|
+ { name: "trail", input: { subject: `size:${REPLY_CAP}` } },
|
||||||
|
+ { name: "trail", input: { subject: `size:${REPLY_CAP + 1}` } },
|
||||||
|
+ { name: "list_tasks", input: {} },
|
||||||
|
+ { name: "get_task", input: { task_ref: "vikunja:3/7" } },
|
||||||
|
+ ]);
|
||||||
|
+ const r = await turn(env, "Message 1 from jason, class REQUEST:\n\nshow the board", s.workspace);
|
||||||
|
+ assert.equal(r.code, 0, r.stderr);
|
||||||
|
+ const results = JSON.parse(r.stdout.trim().slice("ANSWER ".length));
|
||||||
|
+ assert.equal(results.length, 4);
|
||||||
|
+ assert.deepEqual(results[0], [false, `"${"x".repeat(79)}`]);
|
||||||
|
+ assert.equal(results[1][0], true);
|
||||||
|
+ assert.match(results[1][1], new RegExp(`^reply too large: ${REPLY_CAP + 1} characters, over the ${REPLY_CAP}-character cap`));
|
||||||
|
+ assert.deepEqual(results[2], [false, '{"tool":"list_tasks","args":{"done":false}}']);
|
||||||
|
+ assert.deepEqual(results[3], [false, '{"tool":"get_task","args":{"task_ref":"vikunja:3/7"}}']);
|
||||||
|
+ assert.deepEqual(sock.calls.map((c) => [c.tool, c.args]), [
|
||||||
|
+ ["trail", { subject: `size:${REPLY_CAP}` }],
|
||||||
|
+ ["trail", { subject: `size:${REPLY_CAP + 1}` }],
|
||||||
|
+ ["list_tasks", { done: false }],
|
||||||
|
+ ["get_task", { task_ref: "vikunja:3/7" }],
|
||||||
|
+ ]);
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
test("pi: a write through a dangling symlink is blocked, and nothing appears outside", async (t) => {
|
||||||
|
const { dir, workspace, s, env } = await session(t, [
|
||||||
|
{ name: "write", input: { path: "notes.md", content: "planted\n" } },
|
||||||
|
diff --git a/packages/harness/tests/tools.test.mjs b/packages/harness/tests/tools.test.mjs
|
||||||
|
index 3fe7dd1e..045f11c1 100644
|
||||||
|
--- a/packages/harness/tests/tools.test.mjs
|
||||||
|
+++ b/packages/harness/tests/tools.test.mjs
|
||||||
|
@@ -1,6 +1,7 @@
|
||||||
|
import { test } from "node:test";
|
||||||
|
import assert from "node:assert/strict";
|
||||||
|
-import { callTool, typedTools, untypedActions } from "../src/tools.mjs";
|
||||||
|
+import { callTool, REPLY_CAP, replyText, typedTools, untypedActions } from "../src/tools.mjs";
|
||||||
|
+import { TASK_PAGE, TASK_PAGE_MAX } from "../../bus/src/broker.mjs";
|
||||||
|
import { fakeToolSocket, resolvedFor, scratch } from "./helpers.mjs";
|
||||||
|
|
||||||
|
const names = (tools) => tools.map((t) => t.name);
|
||||||
|
@@ -12,7 +13,7 @@ test("the PM gets launch, its task verbs and the reads", (t) => {
|
||||||
|
"send_message", "launch",
|
||||||
|
"task_create", "task_assign", "task_reassign", "task_schedule", "task_priority_change", "task_scope_change",
|
||||||
|
"task_update_assigned", "task_close",
|
||||||
|
- "raise_decision", "resolve_decision", "list_agents", "inbox", "trail", "list_tasks",
|
||||||
|
+ "raise_decision", "resolve_decision", "list_agents", "inbox", "trail", "list_tasks", "get_task",
|
||||||
|
]);
|
||||||
|
const launch = tools.find((x) => x.name === "launch");
|
||||||
|
assert.equal(launch.verb, "launch");
|
||||||
|
@@ -37,7 +38,7 @@ test("a coder with a tracker gets its own task tools and no task_reassign", (t)
|
||||||
|
const coder = resolvedFor(t, "coder");
|
||||||
|
assert.deepEqual(names(typedTools({ resolved: coder, tracker: true })), [
|
||||||
|
"send_message", "task_scope_change", "task_update_assigned",
|
||||||
|
- "raise_decision", "list_agents", "inbox", "trail", "list_tasks",
|
||||||
|
+ "raise_decision", "list_agents", "inbox", "trail", "list_tasks", "get_task",
|
||||||
|
]);
|
||||||
|
});
|
||||||
|
|
||||||
|
@@ -49,7 +50,7 @@ test("launch only when the business's launch block names the instance as launche
|
||||||
|
|
||||||
|
test("an action outside the instance's authority has no tool", (t) => {
|
||||||
|
const coder = resolvedFor(t, "coder", (d) => (d.roles.coder.vars["limits.authority"] = ["review.request"]));
|
||||||
|
- assert.deepEqual(names(typedTools({ resolved: coder, tracker: true })), ["raise_decision", "list_agents", "inbox", "trail", "list_tasks"]);
|
||||||
|
+ assert.deepEqual(names(typedTools({ resolved: coder, tracker: true })), ["raise_decision", "list_agents", "inbox", "trail", "list_tasks", "get_task"]);
|
||||||
|
});
|
||||||
|
|
||||||
|
test("callTool: one JSON line out, the result back, a refusal rejects", async (t) => {
|
||||||
|
@@ -63,3 +64,54 @@ test("callTool: one JSON line out, the result back, a refusal rejects", async (t
|
||||||
|
await assert.rejects(callTool(`${dir}/missing.sock`, "x"), /tool socket: ENOENT/);
|
||||||
|
assert.deepEqual(sock.calls.map((c) => c.tool), ["list_agents", "launch"]);
|
||||||
|
});
|
||||||
|
+
|
||||||
|
+// Row 89 (#1583): task reads a session can page, and replies that fit.
|
||||||
|
+test("list_tasks takes the broker's filters and pages; get_task takes a task_ref", (t) => {
|
||||||
|
+ const tools = typedTools({ resolved: resolvedFor(t, "pm"), tracker: true });
|
||||||
|
+ const list = tools.find((x) => x.name === "list_tasks");
|
||||||
|
+ const get = tools.find((x) => x.name === "get_task");
|
||||||
|
+ assert.deepEqual([list.verb, list.action, get.verb, get.action], ["tasks", null, "tasks", null]);
|
||||||
|
+ const p = list.parameters;
|
||||||
|
+ assert.deepEqual(Object.keys(p.properties), ["done", "assignee", "label", "requirement", "limit", "after"]);
|
||||||
|
+ assert.deepEqual(p.required, []);
|
||||||
|
+ assert.equal(p.properties.done.type, "boolean");
|
||||||
|
+ assert.equal(p.properties.assignee.type, "string");
|
||||||
|
+ assert.deepEqual([p.properties.label.type, p.properties.label.minimum], ["integer", 1]);
|
||||||
|
+ assert.equal(p.properties.requirement.pattern, "^REQ-[A-Z]+-[1-9][0-9]*$");
|
||||||
|
+ assert.equal(p.properties.after.type, "string");
|
||||||
|
+ // The page sizes are the broker's own.
|
||||||
|
+ assert.deepEqual([p.properties.limit.type, p.properties.limit.minimum, p.properties.limit.maximum], ["integer", 1, TASK_PAGE_MAX]);
|
||||||
|
+ assert.match(p.properties.limit.description, new RegExp(`\\b${TASK_PAGE} by default`));
|
||||||
|
+ assert.match(list.description, new RegExp(`open tasks \\(done false\\), ${TASK_PAGE} at a time`));
|
||||||
|
+ assert.match(list.description, /next as after/);
|
||||||
|
+ assert.match(list.description, /get_task/);
|
||||||
|
+ assert.deepEqual(get.parameters.properties, { task_ref: tools.find((x) => x.name === "task_close").parameters.properties.task_ref });
|
||||||
|
+ assert.deepEqual(get.parameters.required, ["task_ref"]);
|
||||||
|
+ // Neither without a tracker.
|
||||||
|
+ assert.deepEqual(names(typedTools({ resolved: resolvedFor(t, "pm") })).filter((n) => n.endsWith("_tasks") || n === "get_task"), []);
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
+test("callTool sends list_tasks done false unless the session names done; other tools as given", async (t) => {
|
||||||
|
+ const sock = await fakeToolSocket(t, scratch(t), (tool, args) => args);
|
||||||
|
+ assert.deepEqual(await callTool(sock.path, "list_tasks"), { done: false });
|
||||||
|
+ assert.deepEqual(await callTool(sock.path, "list_tasks", { limit: 5 }), { done: false, limit: 5 });
|
||||||
|
+ assert.deepEqual(await callTool(sock.path, "list_tasks", { done: true }), { done: true });
|
||||||
|
+ assert.deepEqual(await callTool(sock.path, "get_task", { task_ref: "vikunja:3/7" }), { task_ref: "vikunja:3/7" });
|
||||||
|
+ assert.deepEqual(await callTool(sock.path, "list_agents"), {});
|
||||||
|
+});
|
||||||
|
+
|
||||||
|
+test("replyText: compact JSON up to the cap, an error naming the size and the cap over it", () => {
|
||||||
|
+ assert.equal(REPLY_CAP, 40_000);
|
||||||
|
+ assert.equal(replyText({ a: [1, { b: "c" }] }), '{"a":[1,{"b":"c"}]}');
|
||||||
|
+ assert.equal(replyText(null), "null");
|
||||||
|
+ assert.equal(replyText(undefined), "null");
|
||||||
|
+ // A JSON string adds its two quotes; length counts UTF-16 units, as the CLI does.
|
||||||
|
+ assert.equal(replyText("x".repeat(REPLY_CAP - 2)).length, REPLY_CAP);
|
||||||
|
+ assert.throws(() => replyText("x".repeat(REPLY_CAP - 1)), (e) => {
|
||||||
|
+ assert.equal(e.message, `reply too large: ${REPLY_CAP + 1} characters, over the ${REPLY_CAP}-character cap. Narrow the query: a filter, a smaller limit, or get_task for one task.`);
|
||||||
|
+ return true;
|
||||||
|
+ });
|
||||||
|
+ assert.throws(() => replyText("\u{1F600}".repeat(REPLY_CAP / 2)), /reply too large: 40002 characters/);
|
||||||
|
+ // 40000 units fit, though they are 80002 bytes: the cap isn't a byte count.
|
||||||
|
+ assert.equal(replyText("\u{1F600}".repeat(REPLY_CAP / 2 - 1)).length, REPLY_CAP);
|
||||||
|
+});
|
||||||
@@ -0,0 +1,12 @@
|
|||||||
|
81b985a837d1b5f837fb94faa2264684f6c1633e916e2ab99f57659bc2bd1a9f packages/bus/README.md
|
||||||
|
dbef704b063f04e0ad451cb1977ab2641a9daf37b12537d366d4f22b3299a639 packages/bus/src/broker.mjs
|
||||||
|
7a62dd8d948b576f5873e70c92eee8d51d8fbc69656ff482d30f25ba78d45e31 packages/bus/tests/task-page.test.mjs
|
||||||
|
f334d4c93ec207528775029d3d042922796e5cd950d30d3da55c14977947e525 packages/harness/README.md
|
||||||
|
dab8f6528a836cd8649d46de48b1eb0c06b16a2c0d802e0c22485cd755293812 packages/harness/src/bundle.mjs
|
||||||
|
ce0107c3b21e07925b6543b6b925da8b8768823448d4fcf30b620883cfd2d39e packages/harness/src/mcp-server.mjs
|
||||||
|
fcba4b8dfd9dde947a79f88376d880f383e2a04ee3c0a9ba2fa5af641fe61b35 packages/harness/src/pi-extension.mjs
|
||||||
|
7e8ba55bc5c00274a3c953f0afcde86801a624914f2c5babb97135f7aa561b89 packages/harness/src/tools.mjs
|
||||||
|
414254c1a60c528947ade47388d2a60b91f6964911b737cfeb310e85c97868eb packages/harness/tests/bundle.test.mjs
|
||||||
|
2a84dac2ab5330ef9c96ca1ea878936e09ce589571fe18196c4d2545309cd9d9 packages/harness/tests/mcp-server.test.mjs
|
||||||
|
d1c334a87157d8a8da7f4c592c965b75684dbaac901791341784490fa9b1c2f7 packages/harness/tests/pi-session.test.mjs
|
||||||
|
103b774d6a52f267b0c40f7a944d5a2e3732a2a36fd2c9b8a83d207ee063e1fa packages/harness/tests/tools.test.mjs
|
||||||
@@ -0,0 +1,18 @@
|
|||||||
|
start 2026-10-12T20:13:35Z HEAD-base a3f6e9e66cc1961becb2c4a0ff87fdbdff5f5edc
|
||||||
|
node bus rc=0 ℹ tests 108 ℹ pass 108 ℹ fail 0
|
||||||
|
node harness rc=0 ℹ tests 65 ℹ pass 65 ℹ fail 0
|
||||||
|
node webui rc=0 ℹ tests 84 ℹ pass 84 ℹ fail 0
|
||||||
|
node cli rc=0 ℹ tests 115 ℹ pass 115 ℹ fail 0
|
||||||
|
node tasks rc=0 ℹ tests 74 ℹ pass 74 ℹ fail 0
|
||||||
|
node control-board rc=0 ℹ tests 124 ℹ pass 124 ℹ fail 0
|
||||||
|
test-auth rc=0 selftest: 15 passed, 0 failed
|
||||||
|
test-conductor rc=0 selftest: 17 passed, 0 failed
|
||||||
|
test-config rc=0 selftest: 24 passed, 0 failed
|
||||||
|
test-discord rc=0 discord suite: 66 passed, 0 failed
|
||||||
|
test-extension-package rc=0 extension package selftest: 18 passed, 0 failed
|
||||||
|
test-foundation rc=0 OK differential schema oracle: PASS: differential schema oracle (finite corpus; compatibility evidence, not equivalence proof) selftest: 44 passed, 0 failed
|
||||||
|
test-queue rc=0 queue suite: 27 passed, 0 failed
|
||||||
|
test-release rc=0 selftest: 14 passed, 0 failed
|
||||||
|
test-task rc=0 selftest: 26 passed, 0 failed
|
||||||
|
end 2026-10-12T20:18:12Z
|
||||||
|
rerun 2026-10-12T20:43:56Z to 2026-10-12T20:44:07Z, final candidate (tools.test.mjs gained one assertion after the mutants): node harness rc=0 ℹ tests 65 ℹ pass 65 ℹ fail 0
|
||||||
Executable
+18
@@ -0,0 +1,18 @@
|
|||||||
|
#!/bin/bash
|
||||||
|
# Row 89 gate: sequential, in the candidate worktree, output teed per suite.
|
||||||
|
cd $HOME/dewey-scratch/r89 || exit 9
|
||||||
|
export TMPDIR=$HOME/dewey-scratch/tmp
|
||||||
|
OUT=$HOME/dewey-scratch/r89-out/gate
|
||||||
|
mkdir -p $OUT
|
||||||
|
S=$OUT/../gate-summary.txt
|
||||||
|
echo "start $(date -u +%Y-%m-%dT%H:%M:%SZ) HEAD-base $(git rev-parse HEAD)" > $S
|
||||||
|
for p in bus harness webui cli tasks control-board; do
|
||||||
|
node --test "packages/$p/tests/*.test.mjs" > $OUT/node-$p.log 2>&1; rc=$?
|
||||||
|
echo "node $p rc=$rc $(grep -E '^ℹ (tests|pass|fail)' $OUT/node-$p.log | tr '\n' ' ')" >> $S
|
||||||
|
done
|
||||||
|
for f in scripts/test-*.sh; do
|
||||||
|
n=$(basename $f .sh)
|
||||||
|
if [ "$n" = test-task ]; then DOCKER_HOST=unix:///nonexistent bash $f > $OUT/$n.log 2>&1; else bash $f > $OUT/$n.log 2>&1; fi; rc=$?
|
||||||
|
echo "$n rc=$rc $(grep -iE '[0-9]+ (passed|failed)|pass(ed)?:|fail(ed)?:' $OUT/$n.log | tail -2 | tr '\n' ' ')" >> $S
|
||||||
|
done
|
||||||
|
echo "end $(date -u +%Y-%m-%dT%H:%M:%SZ)" >> $S
|
||||||
@@ -0,0 +1,33 @@
|
|||||||
|
start 2026-10-12T20:18:19Z
|
||||||
|
killed no args filtered too
|
||||||
|
killed tombstones kept
|
||||||
|
killed next on an exact last page
|
||||||
|
killed default limit ignored
|
||||||
|
killed limit max unchecked
|
||||||
|
killed assignee by raw id
|
||||||
|
killed assignees keep persons
|
||||||
|
killed task_ref with other args
|
||||||
|
killed missing task_ref is null
|
||||||
|
killed done check dropped
|
||||||
|
killed done filter dropped
|
||||||
|
killed assignee any string
|
||||||
|
killed label check dropped
|
||||||
|
killed label filter dropped
|
||||||
|
killed requirement pattern dropped
|
||||||
|
killed requirement filter dropped
|
||||||
|
Traceback (most recent call last):
|
||||||
|
File "/home/jwoltje/dewey-scratch/r89-out/mutants.py", line 57, in <module>
|
||||||
|
r = subprocess.run(['node', '--test', *tests], cwd=M, env=env, capture_output=True, text=True, timeout=900)
|
||||||
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
||||||
|
File "/home/jwoltje/.pyenv/versions/3.12.8/lib/python3.12/subprocess.py", line 550, in run
|
||||||
|
stdout, stderr = process.communicate(input, timeout=timeout)
|
||||||
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
||||||
|
File "/home/jwoltje/.pyenv/versions/3.12.8/lib/python3.12/subprocess.py", line 1209, in communicate
|
||||||
|
stdout, stderr = self._communicate(input, endtime, timeout)
|
||||||
|
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
|
||||||
|
File "/home/jwoltje/.pyenv/versions/3.12.8/lib/python3.12/subprocess.py", line 2116, in _communicate
|
||||||
|
self._check_timeout(endtime, orig_timeout, stdout, stderr)
|
||||||
|
File "/home/jwoltje/.pyenv/versions/3.12.8/lib/python3.12/subprocess.py", line 1253, in _check_timeout
|
||||||
|
raise TimeoutExpired(
|
||||||
|
subprocess.TimeoutExpired: Command '['node', '--test', 'packages/bus/tests/*.test.mjs']' timed out after 900 seconds
|
||||||
|
end 2026-10-12T20:34:27Z
|
||||||
@@ -0,0 +1,20 @@
|
|||||||
|
start 2026-10-12T20:35:08Z (run 2: the mutants from 'after ignored' on; run 1 hung there, mutants-run1.txt)
|
||||||
|
killed after ignored (hung: killed after 300 s)
|
||||||
|
killed after inclusive
|
||||||
|
killed total counts the page
|
||||||
|
killed description kept
|
||||||
|
killed tasks keys closed
|
||||||
|
killed ARG_DEFAULTS removed
|
||||||
|
killed default overrides the caller
|
||||||
|
killed cap at >=
|
||||||
|
SURVIVED cap counts bytes
|
||||||
|
killed indented JSON
|
||||||
|
killed cap 50000
|
||||||
|
killed mcp server uncapped
|
||||||
|
killed pi uncapped
|
||||||
|
killed tool limit max 500
|
||||||
|
killed get_task not required
|
||||||
|
killed get_task without tracker
|
||||||
|
killed bundle line pi names
|
||||||
|
killed bundle none dropped
|
||||||
|
end 2026-10-12T20:43:08Z
|
||||||
@@ -0,0 +1,3 @@
|
|||||||
|
start 2026-10-12T20:43:36Z (run 3: the survivor, after the test added in tools.test.mjs)
|
||||||
|
killed cap counts bytes
|
||||||
|
end 2026-10-12T20:43:47Z
|
||||||
@@ -0,0 +1,69 @@
|
|||||||
|
# Row 89 mutants (Dewey): each swaps one exact string in the scratch copy of the candidate, runs
|
||||||
|
# the bus or harness node tests, restores the file, and reports killed (tests fail) or SURVIVED.
|
||||||
|
import subprocess, sys, os, signal
|
||||||
|
M = os.path.expanduser('~/dewey-scratch/r89mut')
|
||||||
|
env = dict(os.environ, TMPDIR=os.path.expanduser('~/dewey-scratch/tmp'))
|
||||||
|
B = 'packages/bus/src/broker.mjs'
|
||||||
|
T = 'packages/harness/src/tools.mjs'
|
||||||
|
MCP = 'packages/harness/src/mcp-server.mjs'
|
||||||
|
PI = 'packages/harness/src/pi-extension.mjs'
|
||||||
|
BU = 'packages/harness/src/bundle.mjs'
|
||||||
|
BUS = ['packages/bus/tests/*.test.mjs']
|
||||||
|
HAR = ['packages/harness/tests/*.test.mjs']
|
||||||
|
muts = [
|
||||||
|
('no args filtered too', B, " if (!Object.keys(a).length) return rows;\n", "", BUS),
|
||||||
|
('tombstones kept', B, " .filter((r) => !r.fields.gone)\n", "", BUS),
|
||||||
|
('next on an exact last page', B, "rest.length > limit ?", "rest.length >= limit ?", BUS),
|
||||||
|
('default limit ignored', B, "const limit = a.limit ?? TASK_PAGE;", "const limit = a.limit ?? TASK_PAGE_MAX;", BUS),
|
||||||
|
('limit max unchecked', B, "!positive(a.limit, TASK_PAGE_MAX)", "!positive(a.limit)", BUS),
|
||||||
|
('assignee by raw id', B, ".filter((u) => bots.has(u)).map((u) => bots.get(u)).sort()", ".sort()", BUS),
|
||||||
|
('assignees keep persons', B, ".filter((u) => bots.has(u)).map((u) => bots.get(u))", ".map((u) => bots.get(u) ?? u)", BUS),
|
||||||
|
('task_ref with other args', B, "Object.keys(a).length !== 1 || ", "", BUS),
|
||||||
|
('missing task_ref is null', B, "?? fail('task-not-found')", "?? null", BUS),
|
||||||
|
('done check dropped', B, "('done' in a && typeof a.done !== 'boolean') ||", "", BUS),
|
||||||
|
('done filter dropped', B, "(!('done' in a) || t.done === a.done) &&", "", BUS),
|
||||||
|
('assignee any string', B, "!(typeof a.assignee === 'string' && Object.hasOwn(roles, a.assignee))", "typeof a.assignee !== 'string'", BUS),
|
||||||
|
('label check dropped', B, "('label' in a && !positive(a.label)) ||", "", BUS),
|
||||||
|
('label filter dropped', B, "(!('label' in a) || t.labels.includes(a.label)) &&", "", BUS),
|
||||||
|
('requirement pattern dropped', B, " && /^REQ-[A-Z]+-[1-9][0-9]*$/.test(a.requirement)", "", BUS),
|
||||||
|
('requirement filter dropped', B, "(!('requirement' in a) || t.requirement === a.requirement)", "true", BUS),
|
||||||
|
('after ignored', B, "matched.filter((t) => t.task_ref > a.after)", "matched", BUS),
|
||||||
|
('after inclusive', B, "t.task_ref > a.after", "t.task_ref >= a.after", BUS),
|
||||||
|
('total counts the page', B, "total: matched.length", "total: tasks.length", BUS),
|
||||||
|
('description kept', B, " title: f.title ?? null,\n", " title: f.title ?? null,\n description: f.description,\n", BUS),
|
||||||
|
('tasks keys closed', B, "verb === 'tasks' ? TASK_ARGS : []", "[]", BUS),
|
||||||
|
('ARG_DEFAULTS removed', T, "args: { ...ARG_DEFAULTS[tool], ...args }", "args", HAR),
|
||||||
|
('default overrides the caller', T, "args: { ...ARG_DEFAULTS[tool], ...args }", "args: { ...args, ...ARG_DEFAULTS[tool] }", HAR),
|
||||||
|
('cap at >=', T, "if (text.length > REPLY_CAP)", "if (text.length >= REPLY_CAP)", HAR),
|
||||||
|
('cap counts bytes', T, "if (text.length > REPLY_CAP)", "if (Buffer.byteLength(text) > REPLY_CAP)", HAR),
|
||||||
|
('indented JSON', T, "const text = JSON.stringify(result) ?? \"null\";", "const text = JSON.stringify(result, null, 2) ?? \"null\";", HAR),
|
||||||
|
('cap 50000', T, "export const REPLY_CAP = 40_000;", "export const REPLY_CAP = 50_000;", HAR + BUS),
|
||||||
|
('mcp server uncapped', MCP, "text: replyText(result)", "text: JSON.stringify(result)", HAR),
|
||||||
|
('pi uncapped', PI, "text: replyText(result)", "text: JSON.stringify(result)", HAR),
|
||||||
|
('tool limit max 500', T, "maximum: TASK_PAGE_MAX,", "maximum: 500,", HAR),
|
||||||
|
('get_task not required', T, "obj({ task_ref: TASK_REF }, [\"task_ref\"])", "obj({ task_ref: TASK_REF })", HAR),
|
||||||
|
('get_task without tracker', T, " add(\"get_task\", \"tasks\",", " }\n {\n add(\"get_task\", \"tasks\",", HAR),
|
||||||
|
('bundle line pi names', BU, "(session.harness === \"pi\" ? resolved.limits.tools : claudeBuiltins({ tools: resolved.limits.tools }))", "resolved.limits.tools", HAR),
|
||||||
|
('bundle none dropped', BU, ".join(\", \") || \"none\"}", ".join(\", \")}", HAR),
|
||||||
|
]
|
||||||
|
only = sys.argv[1:]
|
||||||
|
for name, f, old, new, tests in muts:
|
||||||
|
if only and not any(o in name for o in only): continue
|
||||||
|
p = os.path.join(M, f); src = open(p).read()
|
||||||
|
n = src.count(old)
|
||||||
|
if n != 1: print(f'BAD-ANCHOR ({n}) {name}', flush=True); continue
|
||||||
|
open(p, 'w').write(src.replace(old, new))
|
||||||
|
# Its own process group, so a hung test's children die with it (a mutant that loops forever
|
||||||
|
# never fails on its own; a hang is a kill).
|
||||||
|
proc = subprocess.Popen(['node', '--test', *tests], cwd=M, env=env, stdout=subprocess.PIPE, stderr=subprocess.STDOUT, text=True, start_new_session=True)
|
||||||
|
try:
|
||||||
|
out, _ = proc.communicate(timeout=300)
|
||||||
|
except subprocess.TimeoutExpired:
|
||||||
|
os.killpg(proc.pid, signal.SIGKILL); proc.communicate()
|
||||||
|
out = None
|
||||||
|
finally:
|
||||||
|
open(p, 'w').write(src)
|
||||||
|
if out is None:
|
||||||
|
print(f'killed {name} (hung: killed after 300 s)', flush=True); continue
|
||||||
|
fails = [l for l in out.splitlines() if l.startswith('not ok')]
|
||||||
|
print(('killed ' if proc.returncode else 'SURVIVED') + f' {name}' + (f' ({len(fails)} failing)' if fails else ''), flush=True)
|
||||||
@@ -0,0 +1,16 @@
|
|||||||
|
// Scratch MCP server for row 89: tool `big` returns a text block of exactly n characters.
|
||||||
|
import { createInterface } from "node:readline";
|
||||||
|
const send = (m) => process.stdout.write(JSON.stringify({ jsonrpc: "2.0", ...m }) + "\n");
|
||||||
|
createInterface({ input: process.stdin }).on("line", (line) => {
|
||||||
|
if (!line.trim()) return;
|
||||||
|
const m = JSON.parse(line);
|
||||||
|
if (m.method === "initialize") return send({ id: m.id, result: { protocolVersion: m.params?.protocolVersion ?? "2025-06-18", capabilities: { tools: {} }, serverInfo: { name: "probe", version: "1" } } });
|
||||||
|
if (m.method === "ping") return send({ id: m.id, result: {} });
|
||||||
|
if (m.method === "tools/list") return send({ id: m.id, result: { tools: [{ name: "big", description: "Return n characters", inputSchema: { type: "object", properties: { n: { type: "integer" } }, required: ["n"] } }] } });
|
||||||
|
if (m.method === "tools/call") {
|
||||||
|
const n = m.params.arguments.n;
|
||||||
|
const unit = "abcdefghij";
|
||||||
|
return send({ id: m.id, result: { content: [{ type: "text", text: unit.repeat(Math.ceil(n / 10)).slice(0, n) }] } });
|
||||||
|
}
|
||||||
|
if (m.id !== undefined) send({ id: m.id, error: { code: -32601, message: "nf" } });
|
||||||
|
});
|
||||||
@@ -0,0 +1,7 @@
|
|||||||
|
# Extract of the probe run (claude -p --output-format stream-json, MCP server probe-server.mjs).
|
||||||
|
# One line per tool call: the n asked for, the length of the text the model got back, its start.
|
||||||
|
claude_code_version 2.1.296 model claude-haiku-5-5
|
||||||
|
n=49990 returned_length=49990 starts='abcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghij'
|
||||||
|
n=50000 returned_length=50000 starts='abcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghijabcdefghij'
|
||||||
|
n=50010 returned_length=1315 starts='Error: result (50,010 characters across 1 line) exceeds maximum allowed tokens. Output has been save'
|
||||||
|
total_cost_usd 0.01452308 is_error False
|
||||||
Reference in New Issue
Block a user