Merge remote-tracking branch 'origin/main' into capy/fl2-F1

CHANGELOG.md

@@ -1,5 +1,23 @@
 # Changelog
 
+## [1.91.108.0] - 2026-10-16
+
+**The weekly census's two #3153 findings were in the test fixtures, not in /ship or /review.** The Oct 12 census went red on `ship-exploratory-late-input` and flagged `review-army-simplification` over its session headroom cap. Both are diagnosed from the run's own transcripts (`docs/evals/headroom-diagnosis-2026-10.md`), and each fixture repair lands with a free test that failed first. No skill prose, budget, threshold or detector changed. Reported by the weekly periodic census bot (#3153).
+
+### What this means for you
+
+- **`ship-exploratory-late-input` no longer gets invited to probe before its QA methods are read.** The fixture's scheduling contract told the session to batch "Reads and read-only discovery" together, and the session counted a QA capture as discovery. It went out in the same response as the Step 9 section Read, before any QA section had returned. The contract now says a capture, a deadline guard start or run, or a diagnostic run is a probe, never discovery. The ordering rule itself stays in /ship's own Step 9 STOP, so the case still catches that STOP being lost.
+- **`review-army-simplification` stops spending its budget on a false premise.** Its prompt said the diff was "over 100 lines" when the fixture is 68, so every run reconciled the mismatch, read the whole review skill and checklist, and wrote an unbounded report (142, 91 and 146 s of a 180 s budget on Oct 12; 159 s pooled). It now stages only the Review Army section and the simplification checklist with recorded scope, stats and diff, as the other Review Army cases do. It selects the specialist with the real `/review --simplification` flag and caps the report at 30 lines. The budget and assertions are unchanged.
+
+### If you…
+
+- **…maintain gstack's eval census:** the next census on this revision (or the paid reruns of the two cases) decides both; until then #3153 stays open.
+
+### For contributors
+
+- `test/qa-exploratory-callers.test.ts` replays census 38212265208's native tool order through `callerProbeReadErrors`, with a reordered negative control, and checks the actual runner callback's discovery definition.
+- `test/review-army-budget.test.ts` drives the registered simplification callback with a mocked runner: one pass control and two negative controls. It checks that the premise matches the staged 68-line diff, which files are staged, and the report bound.
+
 ## [1.91.107.0] - 2026-10-16
 
 **The GBRA-40 field list is fixed: install and check agree on one Bun, `rec` approves an autoplan gate in one word, and pregate never sits silent.** Eighteen Capy threads upgraded to v1.91.101.0 and reported 21 problems (`GBRA-40`, October 14). Each gstack item was reproduced with a test that failed first; the plan went through `/autoplan` (two gate revisions, Claude and GPT-6.1 Sol voices per phase) and two independent code reviews. Two items are Capy's (skills not mounted on some crew threads, Drive copies not re-synced) and are handed off, not worked around.

VERSION

@@ -1 +1 @@
-1.91.107.0
+1.91.108.0

agents-digest/gstack-AGENTS.md

@@ -1,4 +1,4 @@
-# gstack digest v1.91.107.0 — regenerate/re-copy after upgrading gstack
+# gstack digest v1.91.108.0 — 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,

docs/evals/headroom-diagnosis-2026-10.md

@@ -96,3 +96,39 @@ declared its target mid-line ("… by the user. Review target (fixed):
 With the target-field fix the same capture is credited at D5 (521 s).
 Replay: `bun test test/detector-corpus-plan-eng-multi-finding-batching.test.ts`.
 No budget or kind change.
+
+## review-army-simplification: 159 s of 180 s in the Oct 12 census (#3153)
+
+The pass-rate gate pooled 159 s of the 180 s session budget (88.1%, over the
+85% cap); census 38212265208's own panel ran 142, 91 and 146 s. Phases from the
+three transcripts (paid slices 25-27):
+
+| Phase (38212265208) | t1 | t2 | t3 |
+|---|---|---|---|
+| Read the extracted review-SKILL.md, checklist, specialist; search for a "## Step 4.5" heading the carved fixture no longer has | 17 s | 15 s | 36 s |
+| Scope and stats commands, analysis, and reconciling the prompt's "over 100 lines" with the 68-line diff | 64 s | 29 s | 13 s, then a subagent (29 s) and re-verification (21 s) |
+| Compose the unbounded report | 45 s | 33 s | 31 s |
+| Final message (each restates the findings and the premise mismatch) | 11 s | 8 s | 11 s |
+
+The fixture, not the skill, carried the extra work. The case now uses the
+staging its siblings use (`stageReviewArmySession`: recorded scope, stats,
+learnings and diff; only `review-army.md` and `simplification.md` staged),
+selects the specialist with the real `/review --simplification` force flag,
+applies the checklist in-session and bounds the report to 30 lines. Budget and
+assertions are unchanged. Free check: the simplification scenarios in
+`test/review-army-budget.test.ts`.
+
+## ship-exploratory-late-input: a probe batched with its Reads (#3153)
+
+Not a headroom case; recorded here because the cause is also fixture work.
+Census 38212265208 failed it with "probe preceded resource read". Capture 001
+went out in the same provider response as the Reads of the caller's Step 8 and
+Step 9 sections, before any QA section was read (the Oct 7 capture `mtb1007a`
+batched it with the exploratory Read). /ship's prose never names the QA capture
+helper; the fixture prompt supplies it, and the fixture's scheduling contract
+(added 2026-10-06) invited batching "Reads and read-only discovery" without
+listing Read results among the dependencies a probe must wait for. The case
+passed on Oct 3 and Oct 5, before that contract. The contract now classifies
+captures, guard starts and runs as probes, never discovery; the ordering rule
+itself stays in /ship's Step 9 STOP, which the case still has to follow unaided.
+Free check: the census 38212265208 replay in `test/qa-exploratory-callers.test.ts`.

package.json

@@ -1,6 +1,6 @@
 {
   "name": "gstack",
-  "version": "1.91.107",
+  "version": "1.91.108",
   "description": "Garry's Stack — Claude Code skills + fast headless browser. One repo, one install, entire AI engineering workflow.",
   "license": "MIT",
   "type": "module",

test/helpers/qa-callers-fixture.ts

@@ -700,7 +700,7 @@ export function qaCallerSessionOptions(fixture: QaCallerFixture, runId: string):
     prompt: `Load gstack's /${fixture.caller} supplied parent phase from caller-${fixture.caller}.md and resume it on the selected working-tree diff against origin/main. This excerpt comes from ${fixture.runtime}/${fixture.caller}/SKILL.md; resolve installed-relative references there, not from the excerpt file or product directory. That path identifies the asset base, not another entrypoint: do not read or invoke the full parent SKILL.md or rerun its preamble. Earlier preamble/branch/base setup is complete; use the existing local origin/main ref without fetch. Earlier-phase asset locators are ${fixture.runtime}/review/checklist.md and ${fixture.runtime}/qa/templates/functional-report-template.md. Read those files directly when referenced; recursive Glob does not follow the installed asset symlinks. Cross-project learnings are configured off in this owned fixture. ${fixture.reviewStart ? `The actual review-start helper already captured REVIEW_START=${fixture.reviewStart} for this unchanged core pass; retain that token. ` : ''} This fixture evaluates only the supplied parent phase, not later publication stages. The fixture owner recorded these read-only observations when this phase began; they are current until an input changes, so use them instead of re-running those commands, and recheck freshness before completion outputs:\n${fixture.startObservations}\n Read README.md for the project contract and commands. Use diagnostic-client commands such as \`bun scripts/probe.ts <literal>\` for exploratory discoveries and their checkpoint evidence. A required \`bun run test\` is separate suite verification: report it as verification, never as a diagnostic observation or checkpoint anchor/target. Use the production evidence helper to publish each diagnostic checkpoint as \`reports/exploration-NNN.json\`, not inside a nested directory; do not transcribe its observed payload. ${fixture.caseId === 'ship-exploratory-plan-checks' ? 'The previously discovered plan is PLAN.md.' : 'No plan file was found.'} There is no remote service and no release publication is authorized. There is no interactive approver; do not invent answers or permission. Keep normal parent decision gates. Before the completion report and each completed:true review record, read HANDOFF.md and reports/HANDOFF.md if present for any concurrent collaborator update, await the results, and compare evidence with current inputs. If a probe's snapshot differs from an earlier probe's, read reports/HANDOFF.md first; it names any collaborator change. A gstack-review-log completed:true record is a completion, not preliminary bookkeeping; a later handoff read cannot validate an earlier completion.\n\nDeadline bookkeeping additionally permits \`bun ${fixture.runtime}/bin/gstack-qa-deadline start ${fixture.cwd}/reports/deadline.json SECONDS [EARLIER_UTC]\`, \`bun ${fixture.runtime}/bin/gstack-qa-deadline status ${fixture.cwd}/reports/deadline.json\`, and \`bun ${fixture.runtime}/bin/gstack-qa-deadline run ${fixture.cwd}/reports/deadline.json -- bun scripts/probe.ts [literal]\`. These are closed literal forms: SECONDS must be positive and at most 300, EARLIER_UTC is the optional caller absolute deadline: use the section clock's Hard deadline UTC, never its Runner entry UTC, reserve-start time or a clock-read time. The child is only the existing diagnostic client with zero or one literal argument. Resolve these exact helper and state paths; do not use variables, another helper, another state file, nested wrappers, scripts, operators or substitutions. Only this helper may create or change reports/deadline.json and its .qa-deadline- temporary files; never use Write/Edit/MultiEdit on those paths. Record the full outer run command in checkpoints and evidence; keep the unchanged child JSON as observed, separate from prefixed guard diagnostics. A completed expired guard-run is not a probe or a pass: retain its unused checkpoint, report not-run coverage and do not restart the deadline. Keep the 12-probe smoke limit. Required suites and explicit plan checks are outside the bounded smoke budget, not permission to reset it.\n\nFunctional evidence uses the same production helper and existing diagnostic client: \`bun ${fixture.runtime}/bin/gstack-qa-evidence capture ${fixture.cwd}/reports NNN --public --deadline ${fixture.cwd}/reports/deadline.json -- bun scripts/probe.ts [literal]\`. These diagnostic receipts are declared public/synthetic, so --public is approved; a fresh three-digit ID is required each time. Explicit plan probes outside the smoke budget may replace --deadline with --timeout-ms 10000; this does not reset or bypass the smoke deadline. After the first capture, put causal intent in the next capture itself by adding \`--after PREV --hypothesis 'causal hypothesis'\` before \`--\` (PREV = the most recent complete capture ID); the helper publishes reports/exploration-NNN.json linking capture PREV's observation to that probe before running it, so one call is both checkpoi
Line truncated
     appendSystemPrompt: `Caller execution scheduling (fixture contract):
 This session has at most 25 assistant turns, including required verification and final artifacts. The command boundary applies to each Bash call, not to the number of independent tool calls in an assistant turn.
-After required clock and approval prerequisites settle, issue independent source Reads and read-only discovery together as separate native tool calls once their paths and inputs are known. For example, after the first clock read, one response can Read the caller excerpt, README.md, every product file and the referenced checklist and templates. Wait for their results before decisions that depend on them.
+After required clock and approval prerequisites settle, issue independent source Reads and read-only discovery together as separate native tool calls once their paths and inputs are known. Read-only discovery is Read, Glob, Grep and the declared inventory, Git and review-read forms; a capture, a deadline guard start or run, or a diagnostic-client run is a probe, never discovery. For example, after the first clock read, one response can Read the caller excerpt, README.md, every product file and the referenced checklist and templates. Wait for their results before decisions that depend on them.
 The completion reserve is for required verification, affected-input revalidation and artifacts, not an earlier deadline. Keep completing required work within the actual remaining deadline; reserve entry alone is not a reason to stop. Use each native probe's snapshot to distinguish current from superseded evidence before deciding which checks still need revalidation.
 Never group diagnostic probes, checkpoint publication with its next probe (a capture carrying --after is one probe, not a group), or any action with the clock/status/approval result it needs. Shell composition remains forbidden outside the declared forms. Preserve every required Read, probe, verification, freshness check, approval and report field; the turn limit does not authorize skipping work or reporting incomplete work as passed.`,
     workingDirectory: fixture.cwd,

test/qa-exploratory-callers.test.ts

@@ -108,6 +108,53 @@ describe('caller native-event observer controls', () => {
       ['exploratory', 'system-functional'].map(id => [id, ready.find(tool => String(tool.input.file_path).endsWith(`/qa/sections/${id}.md`))] as const))).toEqual([]);
   });
 
+  test('census 38212265208: a probe batched with the caller section Read is flagged, and the scheduling contract never calls a probe discovery', async () => {
+    // Native order of census 38212265208 (gate-census-6, ship-exploratory-late-input): provider
+    // message, tool, input; outputs abbreviated. Capture 001 shares a response with the Step 9 section Read.
+    const R = '/home/runner/.cache/gstack-paid-shard-iHlpIM/tmp/qc-cALQYz';
+    const capture = (id: string, extra = '') => `bun ${R}/host/runtime/bin/gstack-qa-evidence capture ${R}/product/reports ${id} --public --deadline ${R}/product/reports/deadline.json ${extra}-- bun scripts/probe.ts`;
+    const replay: Array<[string, string, Record<string, string>, string]> = [
+      ['YvqEApde', 'Read', { file_path: `${R}/host/runtime/ship/sections/plan-completion.md` }, '## Step 8: Plan Completion Audit'],
+      ['YvqEApde', 'Read', { file_path: `${R}/host/runtime/ship/sections/review-army.md` }, '## Step 9: Pre-Landing Review'],
+      ['YvqEApde', 'Bash', { command: `${capture('001')} 3` }, '{"id":"probe-b5ae2d9c-af87-4e8a-9aab-76972f826079","charter":"happy","input":"3"}'],
+      ['w9Gt4kmL', 'Read', { file_path: `${R}/host/runtime/qa/sections/exploratory.md` }, '# Shared exploratory QA'],
+      ['w9Gt4kmL', 'Read', { file_path: `${R}/host/runtime/bin/gstack-qa-evidence` }, '#!/usr/bin/env bun'],
+      ['w9Gt4kmL', 'Bash', { command: `${R}/host/runtime/bin/gstack-review-read` }, 'NO_REVIEWS'],
+      ['w9Gt4kmL', 'Bash', { command: `${R}/host/runtime/bin/gstack-specialist-stats` }, 'SPECIALIST_STATS: 0 reviews analyzed'],
+      ['cba4VKcQ', 'Bash', { command: `${capture('002', "--after 001 --hypothesis 'upper bound 9 should still print 18' ")} 9` }, '{"id":"probe-baf59bfd-ac54-4a4a-8bec-2cd416154bbf","charter":"plan:nine","input":"9"}'],
+      ['cba4VKcQ', 'Read', { file_path: `${R}/host/runtime/qa/sections/scope.md` }, '### Select the surface before setup'],
+      ['cba4VKcQ', 'Read', { file_path: `${R}/host/runtime/qa/sections/system-functional.md` }, '# Functional QA with repository-native tools'],
+    ];
+    const replayTools = (rows: typeof replay) => callerTools(rows.flatMap(([message, name, input, output], i) =>
+      nativeCall(`toolu_${i}`, name, input, output).map((event: any) => event.type === 'assistant'
+        ? { ...event, session_id: 'census-38212265208', message: { ...event.message, id: `msg_${message}` } } : event)));
+    const probeErrors = (tools: ReturnType<typeof callerTools>) => tools.filter(tool => / capture /.test(String(tool.input.command)))
+      .flatMap(probe => callerProbeReadErrors(probe, ['exploratory', 'system-functional']
+        .map(id => [id, tools.find(tool => String(tool.input.file_path).endsWith(`/qa/sections/${id}.md`))] as const)));
+    const tools = replayTools(replay);
+    const section = tools.find(tool => String(tool.input.file_path).endsWith('/ship/sections/review-army.md'))!;
+    expect(tools.find(tool => / capture /.test(String(tool.input.command)))!.messageId).toBe(section.messageId);
+    expect(probeErrors(tools)).toEqual([
+      'probe preceded resource read: exploratory', 'probe preceded resource read: system-functional',
+      'probe preceded resource read: system-functional',
+    ]);
+    const isProbe = ([, , input]: (typeof replay)[number]) => / capture /.test(input.command ?? '');
+    expect(probeErrors(replayTools([...replay.filter(row => !isProbe(row)),
+      ...replay.filter(isProbe).map(([message, ...rest]) => [`${message}-after-reads`, ...rest] as (typeof replay)[number])]))).toEqual([]);
+
+    const fixture = createQaCallerFixture('ship-exploratory-late-input');
+    try {
+      await runQaCaller(fixture, 'free-read-probe-contract', async options => {
+        const contract = options.appendSystemPrompt!;
+        const discovery = contract.match(/Read-only discovery is [^.]*;([^.]*)\./);
+        expect(discovery?.[1]).toMatch(/capture/);
+        expect(discovery?.[1]).toMatch(/guard start/);
+        expect(discovery?.[1]).toMatch(/never discovery/);
+        return { exitReason: 'error_max_turns', transcript: [] } as unknown as SkillTestResult;
+      });
+    } finally { await fixture.close(); fs.rmSync(fixture.root, { recursive: true, force: true }); }
+  });
+
   test.each(['pending-result', 'same-event', 'same-message-id'])('method Read %s cannot authorize a dependent probe', ordering => {
     const observed = evidence();
     const events = observed.result.transcript as any[];

test/review-army-budget.test.ts

@@ -2,6 +2,7 @@ import { expect, test } from 'bun:test';
 import * as fs from 'node:fs';
 import * as os from 'node:os';
 import * as path from 'node:path';
+import { spawnSync } from 'node:child_process';
 
 const ROOT = path.resolve(import.meta.dir, '..');
 
@@ -158,3 +159,62 @@ afterAll(async () => {
     }
   }, 10_000);
 }
+
+// review-army-simplification headroom (#3153): the session's premise matches its staged
+// diff and the fixture stages only the Review Army section and its one checklist.
+test.each(['structure-finding', 'no-structure-finding', 'disavowed-frame'])
+  ('review-army-simplification registered session contract: %s', scenario => {
+    const dir = fs.mkdtempSync(path.join(os.tmpdir(), 'simplification-contract-'));
+    const facts = path.join(dir, 'facts.json');
+    const script = path.join(dir, 'callback.test.ts');
+    fs.writeFileSync(script, `
+import {describe,mock,test} from 'bun:test';
+import * as fs from 'node:fs';
+import * as path from 'node:path';
+import {spawnSync} from 'node:child_process';
+const root=${JSON.stringify(ROOT)},scenario=${JSON.stringify(scenario)},facts=${JSON.stringify(facts)};
+const actual=await import(path.join(root,'test/helpers/session-runner.ts'));
+let capturedRun={};
+mock.module(path.join(root,'test/helpers/e2e-helpers.ts'),()=>({
+  ROOT:root,runId:'simplification-contract',
+  describeIfSelected:(name,ids,body)=>{if(ids.includes('review-army-simplification'))describe(name,body);},
+  testConcurrentIfSelected:(id,body,timeout)=>{if(id==='review-army-simplification')test(id,body,timeout);},
+  logCost(){},createEvalCollector:()=>null,finalizeEvalCollector:async()=>{},
+  recordE2E:(_collector,_name,_suite,result)=>fs.writeFileSync(facts,JSON.stringify({...capturedRun,exitReason:result.exitReason})),
+}));
+mock.module(path.join(root,'test/helpers/session-runner.ts'),()=>({...actual,runSkillTest:async options=>{
+  const cwd=options.workingDirectory;
+  const shortstat=spawnSync('git',['diff','--shortstat','main...HEAD'],{cwd,encoding:'utf8',timeout:5000}).stdout;
+  const staged=spawnSync('git',['ls-files','--others','--exclude-standard'],{cwd,encoding:'utf8',timeout:5000}).stdout.split('\\n').filter(Boolean).sort();
+  capturedRun={prompt:options.prompt,timeout:options.timeout,maxTurns:options.maxTurns,staged,
+    diffLines:[...shortstat.matchAll(/(\\d+) (?:insertion|deletion)/g)].reduce((sum,m)=>sum+Number(m[1]),0)};
+  fs.writeFileSync(path.join(cwd,'review-output.md'),{
+    'structure-finding':'{"category":"native","path":"date_utils.js","line":7} DateFormatter re-implements Intl.DateTimeFormat.',
+    'no-structure-finding':'Simplification: nothing to report.',
+    'disavowed-frame':'{"category":"native"} Lean already. Ship.',
+  }[scenario]);
+  return {exitReason:'success',browseErrors:[],toolCalls:[],transcript:[],output:'Saved.'};
+}}));
+await import(path.join(root,'test/skill-e2e-review-army.test.ts'));
+`);
+    try {
+      const child = spawnSync(process.execPath, ['test', script], {
+        cwd: dir, encoding: 'utf8', timeout: 20_000,
+        env: {PATH: process.env.PATH ?? '', HOME: dir, TMPDIR: dir, NO_COLOR: '1'},
+      });
+      expect(child.status, child.stdout + child.stderr).toBe(scenario === 'structure-finding' ? 0 : 1);
+      expect(fs.existsSync(facts), child.stdout + child.stderr).toBe(true);
+      const observed = JSON.parse(fs.readFileSync(facts, 'utf8'));
+      expect(observed.timeout).toBe(180_000);
+      expect(observed.maxTurns).toBe(20);
+      expect(observed.diffLines).toBe(68);
+      expect(observed.prompt).toContain(`DIFF_LINES: ${observed.diffLines}`);
+      expect(observed.prompt).not.toMatch(/over 100 lines/i);
+      expect(observed.prompt).toContain('/review --simplification');
+      expect(observed.staged).toEqual(['review-army.md', 'review-specialists/simplification.md']);
+      expect(observed.prompt).toContain('at most 30 lines');
+      expect(observed.prompt).toContain('brief acknowledgement');
+    } finally {
+      fs.rmSync(dir, {recursive: true, force: true});
+    }
+  });

test/skill-e2e-review-army.test.ts

@@ -752,6 +752,7 @@ Write ${dir}/review-output.md with only the selection line, the SPECIALIST REVIE
 
 describeIfSelected('Review Army: Simplification activation', ['review-army-simplification'], () => {
   let dir: string;
+  let observations: string;
 
   beforeAll(() => {
     const repo = setupRepo('army-simplification');
@@ -769,24 +770,27 @@ describeIfSelected('Review Army: Simplification activation', ['review-army-simpl
     repo.run('git', ['add', '.']);
     repo.run('git', ['commit', '-m', 'add date utils']);
 
-    copyReviewFiles(dir);
+    observations = stageReviewArmySession(dir, `SCOPE_FRONTEND=false
+SCOPE_BACKEND=true
+SCOPE_PROMPTS=false
+SCOPE_TESTS=false
+SCOPE_DOCS=false
+SCOPE_CONFIG=false
+SCOPE_MIGRATIONS=false
+SCOPE_API=false
+SCOPE_AUTH=false`, ['simplification'], ['simplification']);
   });
 
   afterAll(() => { try { fs.rmSync(dir, { recursive: true, force: true }); } catch {} });
 
   testConcurrentIfSelected('review-army-simplification', async () => {
     const result = await runSkillTest({
-      prompt: `You are in a git repo on a feature branch that adds a JS utility file.
-Read review-SKILL.md for instructions. Also read review-checklist.md.
-The specialist checklists are in review-specialists/ (testing.md, simplification.md, etc.).
+      prompt: `You are the /review parent on branch feature/date-utils, a JS utility change. The caller invoked /review --simplification.
+${reviewArmyScope(observations, 'its Step 4.6 merge')}
+Selection: --simplification force-includes the Simplification specialist; the recorded DIFF_LINES is below its size gate, so only the flag selects it. Dispatch no other specialist.
+The checklist for this capture is review-specialists/simplification.md. Instead of launching a subagent, read it and apply it yourself against the recorded diff.
 
-Skip the preamble, lake intro, telemetry sections.
-Run Step 4.5 (Review Army) only.
-The base branch is main. The diff is over 100 lines, so the Simplification specialist should activate.
-
-For the specialist dispatch, read review-specialists/simplification.md and apply it against the diff.
-
-Write your findings to ${dir}/review-output.md`,
+Write ${dir}/review-output.md with only the selection line and the SPECIALIST REVIEW block with its PR Quality Score and Simplification footer, in at most 30 lines. After saving the report, finish with a brief acknowledgement rather than repeating the findings in the final response.`,
       workingDirectory: dir,
       maxTurns: 20,
       timeout: 180_000,