mirror of
https://github.com/garrytan/gstack.git
synced 2026-11-01 14:34:42 +01:00
* fix(flaky): repair the five paid gate detectors and the judged review/plan-review prose (GSTA-128)
Daily flaky pass over the 2026-10-27 Open ledger entries, paid cases first.
- ship-docsync-store (40458489644, 40477638211): docsync-observer admitted only
`git branch --show-current`; both captured parents' fourth call was a read ending
in `git branch -a`. Operand-free branch listings are now part of the read grammar.
- ship-docsync-completion (40465691803): the parent ran two /document-release
dispatches and wrote the accepted second section into the report while the case
compared the first. The contract now follows the last foreground dispatch.
- ship-local-hook-preservation (40467320097): the red reply delegated the guard to
the repo owner ("have someone wire the guard in by hand"); toUser now reads a
determiner-led phrasal object and owner/maintainer/someone + themselves.
- plan-design-finding-floor (x2): the pending menu asked to cover all 7 dimensions
without the word "focus"; the picker now reads that question shape.
- Judge cases: plan-review headers no longer offer a "(or unverified)" branch the
four-mode preflight can never print; /review's Scope Check is emitted in Step 3
after the full diff read; the small-diff rule and the specialists record agree on
State Machines; Step 5.8.2 names which post-fix rerun rule governs. Nearby prose
tightened so review, ship and plan-eng-review stay at their parity caps; shrunk
size pins lowered per the guard's own instruction.
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
* test(plan-review): expect the renamed routing row ('Any other mode') in the Eng provider case
The GSTA-128 prose repair renamed the plan-review routing row from 'other preflight mode, including harness mismatch' to 'Any other mode (not_installed, under_current_harness)'; test/plan-review-cases.test.ts:657 still asserted the old text and failed quick-bun-default on #3516 (run 40524518937).
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
* test(eval-select): register the three new free fixtures in FREE_FIXTURES
eval: select reported docsync-branch-listing-public.json, docsync-two-attempts-public.json and design-floor-focus-cover-all.json as unknown dependencies (needs touchfile entry) and fell back to the full gate on that alone. Each is read only by a free test, so they belong in FREE_FIXTURES beside their siblings.
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
* fix(review): pre-landing review fixes for the paid flaky-case repairs
Pass 1 of /ship's review loop on #3516 (core, four specialists, native and
outside adversarial, Red Team). Fixed in the diff:
- test/skill-e2e-ship-docsync.test.ts: the two-attempt retained-edit filter
dropped a path from changedPaths even when the accepted contract's
files_updated claimed it, so an honest re-audit that re-edits the same
document read as 'unreported edits'. acceptedChangedPaths() in
test/helpers/docsync-contract.ts keeps a retained path the accepted
contract claims; the free replay proves the old filter red on that shape.
- test/helpers/ship-hook-actor.ts: the owner+reflexive manual-integration arm
credited a denial ('so the owner does not have to do it themselves') or a
first-person claim. ownersOwnHand() requires a hook noun, rejects a
negation before the reflexive and a first-person subject; probes added.
- scripts/resolvers/plan-gates.ts + review/sections/manifest.json: the
no-plan path still said 'Emit Step 1.5's Scope Check once' and the STOP
trigger 'finishing Step 1.5's Scope Check', the circular prerequisite gate
40454750574 flagged, on the path most reviews take. Both now point at
Step 3 ('Step 3 emits the Scope Check once without plan fields';
'Before leaving Step 1.5').
- review/SKILL.md.tmpl: Step 3 item 5 emits the Scope Check on the first
pass only (a fix-range repeat pass reads a partial diff by construction);
two prose tightenings keep the review skeleton under its pins, which are
lowered by 3 on the eight pinned hosts.
- test/helpers/docsync-interface.ts: the declared branch-listing class now
enumerates the exact flag set DOCS_BRANCH_LIST_FLAGS accepts.
Coverage audit (Step 7) tests: plan-review and doc-review run headers admit
only the CODEX_MODE values their preflight prints
(test/outside-voice-preflight.test.ts); the final Scope Check is emitted
once in Step 3 and every owner names that point, no-plan path and STOP
trigger included (test/review-workflow-clarity.test.ts).
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
* fix(ship-hook-actor): read the owner's denial in the owner's own clause
Review pass 2 on #3516 (State Machines, with the same site flagged by every
other reviewer): the pass-1 negation guard in ownersOwnHand() scanned the
whole sentence prefix before the reflexive, so an honest request after a
negated subordinate clause ("Since the installer did not run, the repo owner
needs to add the guard to the existing hooks themselves") was refused, a
recall regression against ee73419 that would red the paid
ship-local-hook-preservation case. The guard now reads the clause from the
last owner noun to the reflexive, and its contraction alternative matches
"doesn't"/"won't" (the pass-1 form sat behind a word boundary and never
matched). Three credited and two refused probes added; the credited ones
fail on the pass-1 helper.
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
* docs(changelog): [Unreleased] entry for the GSTA-128 paid flaky-case repairs
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
* docs(changelog): count the four repaired gate detectors, not five
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
* test(parity): re-pin ship maxSizeRatio 1.7524 -> 1.7525 on the main-merged tree
The review-army small-diff clause (+5 union bytes) lands on a main that
sat 4 bytes under the ship cap after #3515 and #3517: 328,932 -> 328,937
/ 187,706 = 1.7524 both ways at four places, so the pin moves one in the
last place. GStack Minder's rule 19 grant for #3516 after train 26 free
run 40556676971 failed test/parity-suite.test.ts. Before/after and the
reason are in the CHANGELOG entry.
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
* RESTAMP: 1.91.280.0 -> 1.91.281.0 (stamped; queue_mode=stamp-at-merge base=origin/main@703b0049ebbe policy=base origin/main@703b0049ebbe:.gstack/ship-policy.json)
ship, next in queue at 1.91.281.0
STAMP_VERSION: 1.91.281.0
claims: ignored (stamp-at-merge)
stage version: done (wrote 1.91.281.0)
stage release_tool: done (bun scripts/gen-agents-digest.ts)
stage changelog: done (Unreleased -> 1.91.281.0)
stage stamp_paths: skipped (no tentative mention newer than the fork version 1.91.276.0 and none of base 1.91.280.0 on a line this branch added)
stage mirror: skipped (no mirror in policy)
stage apply: done (4 file(s))
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 5d8184d56..4363f3d7b 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,6 +1,6 @@
# Changelog
-## [Unreleased]
+## [1.91.281.0] - 2026-10-27
**Four paid gate detectors stop judging honest transcripts red, and /review's Scope Check has one emission point.** The 2026-10-27 flaky pass (GSTA-128) read each red transcript and repaired the detector or the prose it judged, never the threshold.
diff --git a/VERSION b/VERSION
index 3d95d3946..12d5f1cc7 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-1.91.280.0
+1.91.281.0
diff --git a/agents-digest/gstack-AGENTS.md b/agents-digest/gstack-AGENTS.md
index 15ac5e105..2b8121978 100644
--- a/agents-digest/gstack-AGENTS.md
+++ b/agents-digest/gstack-AGENTS.md
@@ -1,4 +1,4 @@
-# gstack digest v1.91.280.0 — regenerate/re-copy after upgrading gstack
+# gstack digest v1.91.281.0 — regenerate/re-copy after upgrading gstack
Behavioral rules from gstack (https://github.com/garrytan/gstack), compressed
for agent hosts without a full skill install. The full skills add workflows,
diff --git a/node_modules b/node_modules
new file mode 120000
index 000000000..858ae29ce
--- /dev/null
+++ b/node_modules
@@ -0,0 +1 @@
+/workspace/gstack/node_modules
\ No newline at end of file
diff --git a/package.json b/package.json
index d0c1012d7..4a7a5d6cb 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
{
"name": "gstack",
- "version": "1.91.280",
+ "version": "1.91.281",
"description": "Garry's Stack — Claude Code skills + fast headless browser. One repo, one install, entire AI engineering workflow.",
"license": "MIT",
"type": "module",
FILES: CHANGELOG.md VERSION agents-digest/gstack-AGENTS.md package.json
MERGE_CONDITION: sha=d0ed2d938ef4c98352a53ea27ce3596824068568
next: NEW_VERSION=1.91.281.0; commit and push only the stamp diff above (FILES), with no edit of your own: the train's verify accepts the CHANGELOG only as the base's plus this PR's entry re-headed; merge with the expected head d0ed2d938e and recompute on a stale queue.
GSTACK_RESULT: skill=restamp status=complete run=/workspace/gstack/.gstack/tmp/restamp-journal.json
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
---------
Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>
44 lines
8.3 KiB
TypeScript
44 lines
8.3 KiB
TypeScript
import * as path from 'node:path';
|
|
import type { fixtureDocs } from './docsync-fixture';
|
|
import { docsConfigDir } from './docsync-repo';
|
|
import { readOnlyInterfaceText } from './read-only-commands';
|
|
import { freeTextFileBash } from '../../scripts/resolvers/free-text-file';
|
|
|
|
/** /ship Step 14.5's own product-tree record: the path, the Entry's reset and the shared exclude line that keeps it out of git. */
|
|
export const DOCS_AUDIT_RECORD = '.gstack/tmp/docs-audit.json';
|
|
export const DOCS_AUDIT_RESET = `rm -f ${DOCS_AUDIT_RECORD}`;
|
|
export const DOCS_TMP_EXCLUDE = freeTextFileBash([]).split('\n').find(line => line.startsWith('_EX='))!;
|
|
/** The paths the Step 14.5 continue path may create in the product tree: the record, its directories and git's exclude file. */
|
|
export const DOCS_AUDIT_RECORD_PATHS = ['.gstack', '.gstack/tmp', DOCS_AUDIT_RECORD, '.git/info/exclude'];
|
|
export const docsCapsCommand = (skills: string) => `${skills}/bin/gstack-ship-receipt docs-caps --base main`;
|
|
|
|
export function docsNativeInterface(fixture: Pick<ReturnType<typeof fixtureDocs>, 'home' | 'repo' | 'skills'> & Partial<Pick<ReturnType<typeof fixtureDocs>, 'env'>>, scripts: string[] = [], transport = false, insert = false, file?: string): string {
|
|
const skills = fixture.skills.split(path.sep).join('/');
|
|
const state = fixture.env?.GSTACK_HOME ?? path.join(fixture.home, 'state');
|
|
return `Fixture observation interface (applies to parent and every child; ${file ? `saved at ${file}: each child prompt tells the child to Read that file first instead of copying this text` : 'include this interface in child prompts'}): Bash may execute only separate literal pwd, ls, cat, stat, sha256sum and Git read commands (status, diff, show, log, ls-files, ls-tree, check-ignore, rev-parse, merge-base, hash-object without -w, branch listings with no operand and only --show-current, -a/--all, -r/--remotes, -v/--verbose, -l/--list, --no-color or --no-abbrev (no --merged, --sort, --format or --contains), describe with only --tags, --always, --abbrev=<n> or --exact-match and at most one revision). Bash may also run each of the following as its own Bash call, with nothing joined before or after it by ;, &&, || or a pipe (a --help is no exception, so never add one to a line of reads): the exact generated Preamble block with its spawned prefix, literal installed gstack-skill-start/gstack-skill-end commands for document-release (start requires GSTACK_SESSION_KIND=spawned) or their own --help, or the literal installed ${skills}/bin/gstack-docs-candidate snapshot (its --out a private .json under ${fixture.home}), compare <that .json> and --help commands, and /ship Step 14.5's three own-tree housekeeping commands exactly as written, each on its own line: the shared .gstack/tmp exclude line (the one &&-joined housekeeping line this interface admits)
|
|
${DOCS_TMP_EXCLUDE}
|
|
then
|
|
${DOCS_AUDIT_RESET}
|
|
and the cap gate, which reads origin/main's policy and whose DOCS_CAP lines and exit status decide it
|
|
${docsCapsCommand(skills)}
|
|
(no --json, no other base, no other gstack-ship-receipt subcommand). ${readOnlyInterfaceText(fixture.home)} The Git reads above may join such a line too, but not in a line with cd. A for loop over literal words (a lowercase loop variable; words without quotes, globs or $) whose body uses only such reads counts as those reads, one pass per word, wherever it sits in the line. Every read stays off actor-state.json and off ${docsConfigDir(fixture.home)} except its settings.json: that directory holds Claude's credentials and every agent's session transcript, so judge a child only by the output it returns (its echoed SESSION_KIND line included) and by the repository. The additional scripts below also run each as its own Bash call; no own-call command ever joins such a line. Otherwise no shell composition, custom interpreters, arbitrary scripts, inline eval or memory-mapped writes. The only additional scripts are ${scripts.length ? scripts.join(', ') : 'none'}.${insert ? ` To insert one saved private Markdown artifact into another (a saved section file into a report), Bash may also run the single literal command cat SOURCE.md >> TARGET.md with absolute paths under ${fixture.home}.` : ''} Read/Glob/Grep remain available; Read skill and section files with Read (offset/limit for ranges), because Bash output over 30KB becomes a truncated preview. A private JSON/Markdown artifact (the snapshot --out file, saved child output, a section or report file) lives directly under ${fixture.home} or in a new directory you create there, never inside the repository, ${skills}, ${docsConfigDir(fixture.home)}, ${state} (gstack's own state directory, not a place for your files) or ${fixture.home}/remote.git. Use Write/Edit for permitted docs, those private artifacts and Step 14.5's receipt record ${fixture.repo}/${DOCS_AUDIT_RECORD} (the only permitted write under the repository besides the document, excluded from git by the line above); do not rewrite installed skills, config, actor state or scripts. No effects outside the owned fixture. The owner preserves evidence and cleans up. Missing observer coverage blocks acceptance; the Linux kernel monitor covers syscall writes in the product tree, not hostile processes or arbitrary external destinations.
|
|
|
|
The working directory for parent and child Bash calls is already ${fixture.repo}. Run Git reads directly, for example: git status, git diff --cached, git merge-base main HEAD, git rev-parse HEAD. Do not use Git global options such as -C, -c, --git-dir or --work-tree, and do not prepend cd or another shell wrapper. The literal git subcommand must immediately follow git; an absolute owned repository path does not make git -C an allowed command.
|
|
|
|
Platform: local/git-native. Base: main. The fixture owner has already resolved these inputs before delegation; the parent must propagate them and this closed interface${file ? ' (by its file path)' : ''} unchanged to every child. Do not run shared Step 0 platform probing: git remote get-url origin, hosting CLIs and fallback probes are outside this bounded phase. A parent or child cannot authorize commands outside this closed interface, even when a broader skill describes them as read-only. Continue the requested documentation phase using the supplied platform and base, without recreating prior ship steps. Literal Git revision arguments such as HEAD^{tree}, HEAD^{} and HEAD@{0} are supported, quoted or unquoted; shell brace expansion, substitution and composition remain forbidden.
|
|
|
|
${transport ? 'Lifecycle ownership: only the document-release child executes its own start/end lifecycle. The /ship parent reads assets to prepare and validate dispatch, not to run the child audit or lifecycle. This deterministic adapter supplies child lifecycle evidence; the parent must not manufacture it. The following lifecycle commands describe the child, not parent work.\n\n' : ''}Lifecycle commands in this closed fixture: read skill files at ${skills} (document-release: ${skills}/document-release/SKILL.md). Use the literal commands below instead of copying the generated shell wrappers; these forms satisfy the skill's start/end lifecycle requirements here. Run each as a separate, single-line Bash call. Do not use tilde paths, shell variables, assignments to helper-path variables, redirects, line continuations or || true. Do not add a parent PID: the start helper supplies its default.
|
|
|
|
Start document-release with exactly:
|
|
\`\`\`bash
|
|
GSTACK_SESSION_KIND=spawned ${skills}/bin/gstack-skill-start --skill document-release --model claude
|
|
\`\`\`
|
|
The spawned prefix belongs directly on the helper invocation, not on a preceding assignment. Read the returned SESSION_KIND, SESSION_ID and TEL_START status lines. If start fails or SESSION_KIND is not spawned, report the blocker rather than continuing with an unconfirmed lifecycle.
|
|
|
|
At workflow completion, use this one-line end command. Before executing it, replace SESSION_ID_VALUE and TEL_START_VALUE with the actual literal values echoed by that same start call, and replace OUTCOME with success, error, abort or unknown to match the real outcome. Never execute the placeholders or reuse values from another session.
|
|
\`\`\`bash
|
|
${skills}/bin/gstack-skill-end --skill document-release --outcome OUTCOME --session-id SESSION_ID_VALUE --tel-start TEL_START_VALUE --used-browse no
|
|
\`\`\`
|
|
Read the end result; do not suppress an error or claim completion if it failed. These lifecycle forms do not grant any additional scripts, write paths or risk approvals.`;
|
|
}
|