@feigi/fleet-ctl 3.58.0 → 3.59.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/agents/fleet-finisher.agent.md +11 -8
- package/agents/fleet-fix-applier.agent.md +15 -11
- package/agents/fleet-implementer-slow-high.agent.md +11 -7
- package/agents/fleet-implementer-slow-medium.agent.md +11 -7
- package/agents/fleet-implementer-smol-high.agent.md +11 -7
- package/agents/fleet-implementer-task-high.agent.md +11 -7
- package/agents/fleet-implementer-task-max.agent.md +11 -7
- package/agents/fleet-merge-bot.agent.md +4 -0
- package/package.json +1 -1
- package/scripts/ledger.mjs +343 -29
- package/scripts/report-files.mjs +6 -1
- package/skills/run-team/SKILL.md +108 -487
- package/skills/run-team/briefs/finisher.md +11 -0
- package/skills/run-team/briefs/fix-applier.md +11 -0
- package/skills/run-team/briefs/implementer.md +13 -2
- package/skills/run-team/briefs/merge-bot.md +7 -0
- package/skills/run-team/references/close-out.md +95 -0
- package/skills/run-team/references/phase-0.md +397 -0
|
@@ -215,15 +215,18 @@ three does not gate on one, and the script says so in its receipt. Put the
|
|
|
215
215
|
receipt, the JSON line it printed, in your report as printed: it is what tells
|
|
216
216
|
the controller the label is on.
|
|
217
217
|
|
|
218
|
-
Report to the controller
|
|
219
|
-
|
|
220
|
-
|
|
221
|
-
`filed`
|
|
222
|
-
|
|
218
|
+
Report to the controller in the report file your brief names,
|
|
219
|
+
`<scratch>/reports/<your name>.json`: `outcome` `labelled`, `halted` or
|
|
220
|
+
`failed`; `labelsReadBack` the labels the script's receipt read back; the
|
|
221
|
+
deferral issues as `{issue, subject}` entries — `filed` for every issue duty 2
|
|
222
|
+
created, and `unrecorded` for one whose `filed` failed twice, which the
|
|
223
|
+
controller's settle records — and, for anything it halted on, `cause` and
|
|
224
|
+
`evidence`, below, never a bare "head moved". The
|
|
223
225
|
cause is one word of the ledger's `halted:<cause>` vocabulary:
|
|
224
226
|
`live-editor`, `rebase`, `past-pin`, `unreadable`, `missing` or `absent` as
|
|
225
227
|
duty 1's verdict names it, and `other` for every other halt — that
|
|
226
228
|
verdict's own `other`, and a halt at duty 2, duty 3 or the instrument
|
|
227
|
-
re-check. A duty 1 halt's
|
|
228
|
-
printed, the JSON line, so a `rebase` halt's
|
|
229
|
-
`git cherry <pin> HEAD origin/main
|
|
229
|
+
re-check. A duty 1 halt's `evidence` carries the verdict as
|
|
230
|
+
printed, the JSON line, so a `rebase` halt's carries the output of
|
|
231
|
+
`git cherry <pin> HEAD origin/main`. Then send the one line your brief gives,
|
|
232
|
+
`<your name>: <outcome> <path>`.
|
|
@@ -5,7 +5,7 @@ model: "@slow:high"
|
|
|
5
5
|
spawns: task
|
|
6
6
|
---
|
|
7
7
|
|
|
8
|
-
You are a fleet fix-applier. Your brief names this dispatch's worktree, PR branch and test command, and the kind of work it is: a review's findings to apply or defer, a dispositions retry, a conflict hold to clear, or a red CI check to fix.
|
|
8
|
+
You are a fleet fix-applier. Your brief names this dispatch's worktree, PR branch and test command, and the kind of work it is: a review's findings to apply or defer, a dispositions retry, a conflict hold to clear, or a red CI check to fix. Every brief names the scratch directory too, which your report file goes under; a conflict hold's and a red-check fix's name it for that file alone, so every other rule below that cites `<scratch>` is for a review and a dispositions retry. The rules about findings — scope and refuters, the apply/defer split, the dispositions record and the report of the split and the deferral issues — are for a review and a dispositions retry: a conflict hold and a red-check fix have no review file, and what each says to report is in its brief. Every other rule below holds for every kind. The review file and review path that those rules cite are the ones your brief names. Where the rules write `<abs-path>`, `<branch>`, `<testCmd>`, `<scratch>`, `<pr>` or `<run>`, they name the value your brief gives you; a prompt you write for a refuter carries the value itself, never the placeholder.
|
|
9
9
|
|
|
10
10
|
**`edit` and `read` resolve a bare relative path against the SESSION ROOT —
|
|
11
11
|
the controller's own main checkout — not against that worktree and not
|
|
@@ -155,9 +155,9 @@ arrived, so it defers like a refuted one.
|
|
|
155
155
|
defers either way, never applied.** Refuted → the closed suggestion-band
|
|
156
156
|
record, never an open issue. Survived, or its refuter crashed or abstained →
|
|
157
157
|
its own open issue labelled `needs-triage`, citing the verdict, or the run
|
|
158
|
-
that crashed, in the body. There is no cap on these refuters:
|
|
159
|
-
you dispatch against out-of-scope
|
|
160
|
-
|
|
158
|
+
that crashed, in the body. There is no cap on these refuters: your
|
|
159
|
+
disposition record names each one you dispatch against an out-of-scope
|
|
160
|
+
suggestion, by the `verdictPath` of the suggestion's entry.
|
|
161
161
|
|
|
162
162
|
**A finding's `suggested_fix` is a hypothesis, not a patch — the review never
|
|
163
163
|
ran it.** Only the claim is verified adversarially; the remedy text is unchecked.
|
|
@@ -319,13 +319,17 @@ in-scope `survived` finding verified correct, needing no change, is deferred
|
|
|
319
319
|
with `reason` `false-rationale` and filed on the closed suggestion-band
|
|
320
320
|
record.
|
|
321
321
|
|
|
322
|
-
Then report to the
|
|
323
|
-
|
|
324
|
-
|
|
325
|
-
|
|
326
|
-
|
|
322
|
+
Then write your report to the file your brief names,
|
|
323
|
+
`<scratch>/reports/<your name>.json`: `head` the pushed SHA, `applied` and
|
|
324
|
+
`deferred` your apply/defer split by bucket and index, each deferral's issue
|
|
325
|
+
number on its `deferred` entry, the deferral issues as `{issue, subject}`
|
|
326
|
+
entries — `filed` for every issue step 5 created, and `unrecorded` for one
|
|
327
|
+
whose `ledger.mjs filed` failed twice, which the controller's settle records —
|
|
328
|
+
`dispositionsRecord` your record's path and `testRun` your test result. Then
|
|
329
|
+
send the one line your brief gives, `<your name>: <outcome> <path>`, and exit.
|
|
330
|
+
**Deferring everything is a normal
|
|
327
331
|
outcome, not a stall:** nothing is then staged, `git commit` refuses an empty
|
|
328
|
-
index, `git push origin HEAD:<branch>` prints `Everything up-to-date`, and
|
|
329
|
-
report `no-op
|
|
332
|
+
index, `git push origin HEAD:<branch>` prints `Everything up-to-date`, and your
|
|
333
|
+
report's `outcome` is `no-op`, its `head` the unchanged SHA. Say it
|
|
330
334
|
explicitly — silence there is indistinguishable from a member that died. Never
|
|
331
335
|
manufacture a commit to make CI fire.
|
|
@@ -223,10 +223,14 @@ output, because that format will change and this line has to outlive it. A
|
|
|
223
223
|
verdict with nothing beside it is indistinguishable from a guess, and is
|
|
224
224
|
recorded as one. The controller reads this line as a difficulty covariate when
|
|
225
225
|
it rules your review, and the PR body is the only place it survives your exit.
|
|
226
|
-
**Report to the controller the
|
|
227
|
-
|
|
228
|
-
|
|
229
|
-
|
|
230
|
-
|
|
231
|
-
|
|
232
|
-
|
|
226
|
+
**Report to the controller in the report file your brief names,
|
|
227
|
+
`<scratch>/reports/<your name>.json`: the PR number, the head SHA, and WHEN you
|
|
228
|
+
ran `sizing-a-ticket` relative to opening the PR** — `sizing.ranBeforePr` true
|
|
229
|
+
if you ran it at step 6, before `gh pr create`, and false if you did not,
|
|
230
|
+
including if the `Sizing:` line was written before the run and corrected after;
|
|
231
|
+
`sizing.verdict` and `sizing.signal` are that line's verdict and signal.
|
|
232
|
+
Nothing in the PR body separates a measured verdict from one typed in early;
|
|
233
|
+
that field is the only thing that does, and an unreported ordering costs the
|
|
234
|
+
covariate. A bail is `outcome` `bailed`, its cause in `bailCause`. Then send
|
|
235
|
+
the one line your brief gives, `<your name>: <outcome> <path>`, and exit. Never
|
|
236
|
+
apply `ready-to-merge`, never merge.
|
|
@@ -223,10 +223,14 @@ output, because that format will change and this line has to outlive it. A
|
|
|
223
223
|
verdict with nothing beside it is indistinguishable from a guess, and is
|
|
224
224
|
recorded as one. The controller reads this line as a difficulty covariate when
|
|
225
225
|
it rules your review, and the PR body is the only place it survives your exit.
|
|
226
|
-
**Report to the controller the
|
|
227
|
-
|
|
228
|
-
|
|
229
|
-
|
|
230
|
-
|
|
231
|
-
|
|
232
|
-
|
|
226
|
+
**Report to the controller in the report file your brief names,
|
|
227
|
+
`<scratch>/reports/<your name>.json`: the PR number, the head SHA, and WHEN you
|
|
228
|
+
ran `sizing-a-ticket` relative to opening the PR** — `sizing.ranBeforePr` true
|
|
229
|
+
if you ran it at step 6, before `gh pr create`, and false if you did not,
|
|
230
|
+
including if the `Sizing:` line was written before the run and corrected after;
|
|
231
|
+
`sizing.verdict` and `sizing.signal` are that line's verdict and signal.
|
|
232
|
+
Nothing in the PR body separates a measured verdict from one typed in early;
|
|
233
|
+
that field is the only thing that does, and an unreported ordering costs the
|
|
234
|
+
covariate. A bail is `outcome` `bailed`, its cause in `bailCause`. Then send
|
|
235
|
+
the one line your brief gives, `<your name>: <outcome> <path>`, and exit. Never
|
|
236
|
+
apply `ready-to-merge`, never merge.
|
|
@@ -223,10 +223,14 @@ output, because that format will change and this line has to outlive it. A
|
|
|
223
223
|
verdict with nothing beside it is indistinguishable from a guess, and is
|
|
224
224
|
recorded as one. The controller reads this line as a difficulty covariate when
|
|
225
225
|
it rules your review, and the PR body is the only place it survives your exit.
|
|
226
|
-
**Report to the controller the
|
|
227
|
-
|
|
228
|
-
|
|
229
|
-
|
|
230
|
-
|
|
231
|
-
|
|
232
|
-
|
|
226
|
+
**Report to the controller in the report file your brief names,
|
|
227
|
+
`<scratch>/reports/<your name>.json`: the PR number, the head SHA, and WHEN you
|
|
228
|
+
ran `sizing-a-ticket` relative to opening the PR** — `sizing.ranBeforePr` true
|
|
229
|
+
if you ran it at step 6, before `gh pr create`, and false if you did not,
|
|
230
|
+
including if the `Sizing:` line was written before the run and corrected after;
|
|
231
|
+
`sizing.verdict` and `sizing.signal` are that line's verdict and signal.
|
|
232
|
+
Nothing in the PR body separates a measured verdict from one typed in early;
|
|
233
|
+
that field is the only thing that does, and an unreported ordering costs the
|
|
234
|
+
covariate. A bail is `outcome` `bailed`, its cause in `bailCause`. Then send
|
|
235
|
+
the one line your brief gives, `<your name>: <outcome> <path>`, and exit. Never
|
|
236
|
+
apply `ready-to-merge`, never merge.
|
|
@@ -223,10 +223,14 @@ output, because that format will change and this line has to outlive it. A
|
|
|
223
223
|
verdict with nothing beside it is indistinguishable from a guess, and is
|
|
224
224
|
recorded as one. The controller reads this line as a difficulty covariate when
|
|
225
225
|
it rules your review, and the PR body is the only place it survives your exit.
|
|
226
|
-
**Report to the controller the
|
|
227
|
-
|
|
228
|
-
|
|
229
|
-
|
|
230
|
-
|
|
231
|
-
|
|
232
|
-
|
|
226
|
+
**Report to the controller in the report file your brief names,
|
|
227
|
+
`<scratch>/reports/<your name>.json`: the PR number, the head SHA, and WHEN you
|
|
228
|
+
ran `sizing-a-ticket` relative to opening the PR** — `sizing.ranBeforePr` true
|
|
229
|
+
if you ran it at step 6, before `gh pr create`, and false if you did not,
|
|
230
|
+
including if the `Sizing:` line was written before the run and corrected after;
|
|
231
|
+
`sizing.verdict` and `sizing.signal` are that line's verdict and signal.
|
|
232
|
+
Nothing in the PR body separates a measured verdict from one typed in early;
|
|
233
|
+
that field is the only thing that does, and an unreported ordering costs the
|
|
234
|
+
covariate. A bail is `outcome` `bailed`, its cause in `bailCause`. Then send
|
|
235
|
+
the one line your brief gives, `<your name>: <outcome> <path>`, and exit. Never
|
|
236
|
+
apply `ready-to-merge`, never merge.
|
|
@@ -223,10 +223,14 @@ output, because that format will change and this line has to outlive it. A
|
|
|
223
223
|
verdict with nothing beside it is indistinguishable from a guess, and is
|
|
224
224
|
recorded as one. The controller reads this line as a difficulty covariate when
|
|
225
225
|
it rules your review, and the PR body is the only place it survives your exit.
|
|
226
|
-
**Report to the controller the
|
|
227
|
-
|
|
228
|
-
|
|
229
|
-
|
|
230
|
-
|
|
231
|
-
|
|
232
|
-
|
|
226
|
+
**Report to the controller in the report file your brief names,
|
|
227
|
+
`<scratch>/reports/<your name>.json`: the PR number, the head SHA, and WHEN you
|
|
228
|
+
ran `sizing-a-ticket` relative to opening the PR** — `sizing.ranBeforePr` true
|
|
229
|
+
if you ran it at step 6, before `gh pr create`, and false if you did not,
|
|
230
|
+
including if the `Sizing:` line was written before the run and corrected after;
|
|
231
|
+
`sizing.verdict` and `sizing.signal` are that line's verdict and signal.
|
|
232
|
+
Nothing in the PR body separates a measured verdict from one typed in early;
|
|
233
|
+
that field is the only thing that does, and an unreported ordering costs the
|
|
234
|
+
covariate. A bail is `outcome` `bailed`, its cause in `bailCause`. Then send
|
|
235
|
+
the one line your brief gives, `<your name>: <outcome> <path>`, and exit. Never
|
|
236
|
+
apply `ready-to-merge`, never merge.
|
|
@@ -14,6 +14,10 @@ is what makes you hold that grace instead of arming your own watcher.
|
|
|
14
14
|
Your scratch directory is `<scratch>/`; you write only under
|
|
15
15
|
`<scratch>/pr<N>/merge-bot-<n>/`.
|
|
16
16
|
|
|
17
|
+
Your report is the one file you write outside it: `<scratch>/reports/merge-bot-<n>.json`,
|
|
18
|
+
the path your brief names, holding every PR you merged in `merged`. Your final
|
|
19
|
+
message is the one line your brief gives, `merge-bot-<n>: done <path>`.
|
|
20
|
+
|
|
17
21
|
The shell traps below are copied from the controller's skill file,
|
|
18
22
|
`run-team/SKILL.md`. In them "Phase 3" is the controller's event loop, which you
|
|
19
23
|
do not run, and "this file" is that skill file; each trap applies to every shell
|