Files
gstack/test/helpers/auq-native-capture.ts
Garry Tanandcapy-ai[bot] 3e4240a76f v1.91.276.0 feat(preamble): tier 2+ skills load the shared preamble once per session (GSTA-96) (#3434)
* fix(preamble): completeness means complete coverage with minimal code (GSTA-96)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* feat(preamble): tier 2+ skills load the shared preamble once per session (GSTA-96)

Every tier 2+ SKILL.md inlined the same ~23 KB preamble. It now renders once
per host into sections/preamble.md beside each tier 2+ SKILL.md; the SKILL.md
keeps a pointer that asks for the Read only when the file's id is not already
in context. Only sections byte-identical to the shared render move, so every
skill keeps exactly its pre-hoist paragraphs (verified across all 41 skills).

Parity caps for ten skills raised to their measured unions (owner-approved,
Garry 2026-10-25: "do it"); 63 size pins lowered and 26 deleted.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix(paid-select): a generated sections/preamble.md takes its skill template's identity in the PR planner (GSTA-96)

screen-summary failed on #3434 with 'PR profile requires full validation' for nine
skills whose only paid coverage is periodic: their new sections/preamble.md had no
dependency identity, while the SKILL.md beside it inherits its template's.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test(section-manifest): the shared sections/preamble.md renders from the hoist resolver, not a section template (GSTA-96)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes (CHANGELOG says 3 characters shorter; model-overlays guards the coverage half)

The completeness section shrank by 3 characters (5 bytes), matching the 89 pins lowered by 3. The default-copy test pinned only the minimal-code half; an expectMentions check now fails if the full-coverage or never-excuses-a-gap clause is dropped.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes (model-overlays binds smaller diff to equal coverage)

The default-copy check now also fails if 'on equal coverage' is dropped from the smaller-diff rule (which would tell models to prefer the smaller diff unconditionally) or if the no-impossible-state-checks clause is removed.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: CHANGELOG lists the model-overlays coverage checks

The [Unreleased] Tests bullet now says the default-copy test checks both halves of the rule (full coverage of the requirement; smaller diff only at equal coverage; no impossible-state checks; never excuses a gap), matching 88e3cc0 and 44eacaf.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: plan-ceo-review and ETHOS say completeness means coverage, not more code

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes (CEO bullet bounds code by accepted scope; tests bind each completeness rule)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: cover the shared-preamble hoist (ship Step 7)

Adds test/preamble-hoist.test.ts (tier 1 unchanged; tier 2+ SKILL.md plus
sections/preamble.md reproduces the pre-hoist preamble; pointer position and id)
and two gen-skill-docs-checks cases (one shared preamble per host, beside every
pointer and named by its id; preamble artifact kind).

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: pre-landing review fixes (ship Step 9.4)

- setup: follow a symlinked sections/ dir in the carved-pointer status check
- touchfiles: select preamble-hoist.ts alongside every preamble.ts entry
- section-manifest: exempt sections/preamble.md only for skills whose SKILL.md
  carries the shared-preamble pointer
- skill-size-budget: growth ceiling measures SKILL.md plus its shared preamble
- host-conformance: tolerate a missing sections/ dir, as setup does
- skill-validation: read bin/gstack-skill-start directly
- skill-fixture: error messages and docblocks name the actual boundary
  (harness-version non-behavioral note updated)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: describe the size-budget growth check's startup measure (ship Step 9.4, pass 2)

The growth ceiling now measures SKILL.md plus the shared preamble, the same
measure as the shrink floor; the docblock said SKILL.md only.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: CHANGELOG says simplification.md quotes the old rule and CLAUDE.md restates the new one

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: CHANGELOG lists the hoist tests and the growth-ceiling measure (ship Step 9.4)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: cover the symlinked sections/ status check and the ETHOS house-rule quote (ship Step 7)

host-conformance: --status still finds a broken bare sections/ pointer when a
carved Claude skill's installed sections/ is a directory symlink (fails at the
pre-repair 0e5d11fd9). gen-skill-docs: the house rule review/specialists/
simplification.md quotes from ETHOS.md still appears in ETHOS.md.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: describe the shared preamble file (ship Step 14.5 docs audit)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: the shared-preamble skip needs the skill's own Read; digest and first-run intro say complete coverage, minimal code (GSTA-96)

The pointer skipped the shared file when any text carrying its id was in context, so a
page or file with that heading could stand in for the anti-spoofing rules (security,
red team and adversarial review, ship run 2). It now requires the skill's own earlier
Read. The digest's Boil the Ocean bullet and the first-run intro still said 'AI makes
completeness cheap, so do the complete thing' (docs audit blocker); both now carry
Garry's 'complete coverage, minimal code' (2026-10-25: '1/ yes fix this').

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: pin the shared-preamble provenance skip and the new Boil the Ocean wording (ship Step 7)

preamble-hoist: on every host, tier and render option the pointer allows the
skip only when the skill's own Read is in context. agents-digest and
gstack-skill-start: the digest bullet and the lake intro say complete
coverage, minimal code, never the retired phrasing. Each fails at 9af36d010.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: describe the provenance skip where the docs still described the id-only one (ship Step 9.4)

The preamble-hoist.ts docblock and ARCHITECTURE.md still said the shared
preamble is re-read only when its id is absent; since 2acf1a44a the pointer
requires the skill's own earlier Read of the file.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: CHANGELOG lists the provenance-pointer and Boil the Ocean wording tests (ship Step 13)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* revert(digest): keep the agents digest at main's bytes; an unstamped head cannot change a release output (GSTA-96)

CI 'Check VERSION is not stale vs queue' classified #3434 as stamped because it touched
agents-digest/gstack-AGENTS.md (a release_outputs path), while regenerating only the
generator fails the digest freshness test. The first-run intro rewording stays.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: voice judge reads the hoisted preamble; register ETHOS.md on its reader cases

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: refresh the claude ship golden the main merge took from main (ship Step 5 preflight)

The 1a4dcaf07 merge resolved test/fixtures/golden/claude-ship-SKILL.md to
main's pre-hoist bytes. The golden is a copy of ship/SKILL.md; regenerated
with bun scripts/gen-ship-goldens.ts (pregate regen: REGEN_STALE).

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: pr-dependencies' unplaceable-file example uses SECURITY.md now that ETHOS.md is mapped

6183ca8ae registered ETHOS.md on its reader cases in touchfiles-data.ts, so a
PR touching only ETHOS.md plans in pr mode and no longer restores the full gate.
The fail-closed test and the real-unknown label check now use SECURITY.md, which
is tracked and unmapped; the assertions are unchanged. CHANGELOG names the
third test file. (CI free shard 7 at 7c21720ac; remote pregate shard 6.)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: host path-leak checks read the moved shared preamble too; name SHARED_PREAMBLE_FILE (ship review pass 2)

The external-host path-leak assertions in gen-skill-docs.test.ts (Codex,
Factory) and host-config.test.ts (Copilot, Antigravity) read SKILL.md
alone, so a Claude path that moved into sections/preamble.md with the
hoist went unchecked. They now read the skill's startup text through
readSkillStartup. workflow-judge-input and host-conformance use the
SHARED_PREAMBLE_FILE constant instead of a bare 'preamble.md' literal,
and preamble.ts's header describes the hoisted composition.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: CHANGELOG token savings name the per-skill skip and the autoplan follow-up (ship review pass 2)

The pointer's skip is bound to the skill's own copy of sections/preamble.md,
so "every later one saves about 5,800" holds per skill, not per session, and
/autoplan's skip list and the inline /office-hours loader do not yet skip the
pointer section. The entry now says so instead of claiming the full /autoplan
saving. Adding the skip entry grows size-pinned skills, so it is a follow-up.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: corpus budget measures startup bytes; carve-safety counts shared preamble as uncarved (ship review pass 2)

skill-size-budget's growth ceiling and shrink floor already measured
SKILL.md plus sections/preamble.md, but the total-corpus check still summed
hoisted SKILL.md bytes against the pre-hoist baseline, hiding about 0.9 MB
of headroom. It now sums the same startup bytes (3,114,885 -> 2,590,484,
x0.832); the unreachable SKILL.md-missing fallback is gone.

auq-format-always-loaded's carve-safety guard treated any sections/ dir as
carved, so every skill with only the shared preamble ran it and its sanity
check passed on preamble.md. It now uses setup --status's rule: a section
other than the shared preamble. Those skills keep the per-skill checks.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: remaining SKILL.md-only negatives read the shared preamble too (ship review pass 3)

The parameterized external-host path-leak smoke test, the two Codex
.claude path checks and both vestigial-handshake negatives read SKILL.md
alone, so text that moved into sections/preamble.md went unchecked. They
now read readSkillStartup. preamble.ts's header no longer says tier 1
renders through generatePreamble().

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: ratchet skeleton caps and eager context ceilings down after the hoist (ship review pass 3)

The hoist moved about 22 KB out of every tier 2+ SKILL.md, which left
each hoisted maxSkeletonBytes cap and each eagerPerInvocation ceiling
roughly 22 KB / 5-6K tokens slack. The 20 hoisted skeleton caps now sit at
the post-hoist skeleton plus their pre-hoist headroom, and 42 context-budget
ceilings take the recaptured (lower) values. Nothing was raised: the four
ceilings the recapture would have raised keep their old values.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: preamble-hoist's section splitter closes a fence only on a bare delimiter (ship Step 11)

Inside an open fence, a line such as "```bash" is content in CommonMark,
but splitSections treated it as the closing fence, so a later "## " line in
the same block could split a section mid-code-block and hoist half of it.
A closing delimiter now needs nothing but whitespace after it. Every host's
render is unchanged (gen-skill-docs --dry-run FRESH on all twelve hosts).

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: skill-startup.ts is a global, pinned harness file like skill-fixture.ts (ship review pass 3)

skill-fixture.ts is a global touchfile pinned in scripts/harness-version.json
because it reshapes the skill content most E2E suites read, and it now
builds that content through test/helpers/skill-startup.ts. That helper was
in neither list, so a later edit to it would change fixture inputs without
the harness-version guard and select only 27 keys. It is now global and
pinned, recorded non-behavioral (its fixtures are byte-identical to the
pre-hoist output). ARCHITECTURE.md and the CHANGELOG stop calling the
pointer id a content id: it hashes the shared text before host rewrites.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: CHANGELOG matches the merged skip lists and the baseline diff (ship review, run 4)

The headline still called the autoplan and inline /office-hours skip
entries a follow-up, but the main merge added them; it now says the loaded
skills skip their shared Read. The #3451 sentence is dropped, because that
fix shipped in v1.91.237.0 and is recorded there. The typecheck baseline
line says the TS2769 entry was re-keyed for the 'preamble' artifact kind
with its count unchanged, not fixed and lowered.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test: pin the pointer entry in the loader skip lists; drop a carve-safety tautology (ship review, run 4)

A new INVOKE_SKILL test checks that plan-ceo-review's inline /office-hours
skip list and autoplan's section skip list both name the shared-preamble
pointer heading, and that office-hours and the four plan reviews carry it.
The carve-safety "sanity" assertion repeated the predicate discoverSkills()
now uses to set sectionsDir, so it could never fail; it is removed.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* fix: gstack-doctor finds the Codex model patch in the hoisted shared preamble (ship Step 11, run 4)

The doctor's patch row grepped only the Codex review SKILL.md for the
"Model-Specific Behavioral Patch (" heading. The hoist moves that section
into sections/preamble.md, so the required row (for ship, review, the plan
reviews, office-hours, investigate and more) would fail on every correct
post-hoist Codex install and block install.sh's "gstack: ok". It now checks
the SKILL.md and then its shared preamble and names the file it found. A
new doctor test covers a hoisted render.

preamble-hoist's variantSubsections also returns null when no subsection
differs, instead of emitting a replacing heading with no subsections.
Every host's render is unchanged.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test(docsync): fixtureDocs installs the generated sections/preamble.md (hunk taken from #3472)

A generated skill that loads the shared preamble names sections/preamble.md
as a required asset before skill start, and a spawned document-release child
blocks on its absence, so docsync-spawned stayed red on every #3434 screen.
fixtureDocs now copies that file when the generated skill has it.

This single hunk is taken unchanged from gstack#3472 (head 2e3f7ba1,
GSTA-107), which trains ahead of #3434, so the train merge sees identical
lines. Nothing else from #3472 is included.

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: CHANGELOG notes the docsync fixture hunk taken from #3472

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs: CHANGELOG names the post-merge context-budget ratchet (captain hoisted)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test(auq): planted skill fixtures resolve the shared preamble like an install (GSTA-96)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test(plan-format): mode menu delivers its question without output-harvesting wording (GSTA-96)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(changelog): record the v1.91.270.0 merge values (GSTA-96)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* test(auq): parallel-capture contract expects the planted skill with its sections path mapped (GSTA-96)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(changelog): record the v1.91.273.0 merge re-measurements (GSTA-96)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(changelog): restore the released entries the previous commit truncated (GSTA-96)

Co-authored-by: capy-ai[bot] <230910855+capy-ai[bot]@users.noreply.github.com>

* docs(changelog): end two sentences the prose gate flags (GSTA-96)

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>
2026-10-27 13:28:18 +00:00

398 lines
22 KiB
TypeScript
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
/** Capture an actual public AskUserQuestion, without reading model transcripts. */
import * as fs from 'node:fs';
import * as os from 'node:os';
import * as path from 'node:path';
import { randomUUID } from 'node:crypto';
import { resolveEvalModel } from '../../lib/eval-model';
import { launchClaudePty, capturePlanCountQuestion, parseNumberedOptions, type ClaudePtySession } from './claude-pty-runner';
import { API_ERROR_PANEL } from './pty/screen';
import { buildSeedConfig, isHermeticEnabled } from './hermetic-env';
import { getProjectEvalDir } from './eval-store';
import { readFirstPendingQuestionForDisplay, pendingQuestionRecorderStatus } from './plan-count-pending-question';
import type { NativePlanQuestion, NativePlanQuestionCall } from './plan-count-transcript';
export const NATIVE_AUQ_CAPTURE_MS = 240_000;
// close() reserves at most two seconds for SIGINT and one for SIGKILL.
const CLEANUP_MS = 3_000;
export interface NativeAuqCaptureOptions {
planDir: string;
skillName: string;
scenario: string;
testName: string;
runId?: string;
model?: string;
}
export interface NativeAuqCapture {
outcome: 'question_captured';
workflowCompleted: false;
source: 'pre_tool_use';
sessionId: string;
toolUseId: string;
questionIndex: number;
question: NativePlanQuestion;
publicCall: NativePlanQuestionCall;
text: string;
artifactDir?: string;
}
/** Neutral separators only: every graded word must come from this one question. */
export function serializeNativeAuq(question: NativePlanQuestion): string {
return [question.header, question.question,
...question.options.map(option => [option.label, option.description].filter(value => value !== undefined).join('\n')),
].join('\n\n');
}
/** Counting tolerates damaged labels; format capture must reject visible contradictions. */
function nativeOptionLabelsAgree(screen: string, question: NativePlanQuestion): boolean {
const rows = screen.replace(/\r+\n?/g, '\n').split('\n');
const cursorRow = rows.findLastIndex(row => /❯\s*1\./.test(row));
if (cursorRow < 0) return false;
const menu = rows.slice(cursorRow);
const width = (text: string) => Bun.stringWidth(text);
const left = (text: string, columns: number) => {
let result = '', used = 0;
for (const char of text) {
const next = width(char);
if (used + next > columns || used >= columns) break;
result += char; used += next;
}
return result;
};
// Strip only an actual aligned preview box, not a literal bar inside a label.
let previewColumn: number | undefined;
for (let top = 0; top < menu.length; top++) {
const box = /┌─+┐\s*$/.exec(menu[top]!);
if (!box || box.index === 0) continue;
const column = width(menu[top]!.slice(0, box.index));
const bottom = menu.findIndex((row, i) => i > top &&
row.slice(left(row,column).length).trim() === box[0].trim().replace('┌','└').replace('┐','┘'));
if (bottom > top + 1 && menu.slice(top + 1,bottom).every(row =>
/^│[^\n]*│\s*$/.test(row.slice(left(row,column).length)))) {
previewColumn = column; break;
}
}
const labelRows = menu.map(row => {
if (previewColumn === undefined) return row;
const prefix = left(row,previewColumn), sidebar = row.slice(prefix.length);
return /^(?:[┌│└]|Notes: press n to add notes\s*$)/.test(sidebar) ? prefix.trimEnd() : row;
});
const options = parseNumberedOptions(labelRows.join('\n'));
const optionRow = /^[ \t]{0,3}(?:❯\s*)?([1-9])\.\s*(.*)$/;
// The shared parser requires two choices. A single remaining visible choice
// can still explicitly contradict its indexed native label.
for (const row of labelRows) {
const match = optionRow.exec(row);
if (match && !options.some(option => option.index === Number(match[1]))) {
options.push({index:Number(match[1]),label:match[2]!.trim()});
}
}
const compact = (text: string) => text.replace(/\s+/g,'');
const label = (text: string) => compact(question.multiSelect ? text.replace(/^\[[ ✓✔xX]\]\s*/,'') : text);
return options.every(option => {
const expected = question.options[option.index - 1];
if (!expected) return /^(?:Typesomething\.?|Chataboutthis)$/.test(compact(option.label));
let shown = label(option.label), wanted = compact(expected.label);
if (!shown || shown === wanted) return true;
if (/(?:…|\.\.\.)$/.test(shown)) {
const prefix = shown.replace(/(?:…|\.\.\.)$/,'');
return !prefix || wanted.startsWith(prefix);
}
if (!wanted.startsWith(shown)) return false;
const start = labelRows.findIndex(row => {
const match = optionRow.exec(row);
return match && Number(match[1]) === option.index;
});
if (start < 0) return false;
let last = start;
for (let i = start + 1; i < labelRows.length && /^[ \t]{4,}\S/.test(labelRows[i]!); i++) {
const next = compact(labelRows[i]!.trim());
if (!wanted.startsWith(shown + next)) return false;
shown += next; last = i;
if (shown === wanted) return true;
}
// Prefix-only equality requires visible clipping at the actual 120-column
// viewport edge; a shorter complete row such as Keep != Keep current fails.
return width(menu[last]!.trimEnd()) === 120 && wanted.startsWith(shown);
});
}
/** The native UI can clip a tall pane's top and elide its tail simultaneously. */
function clippedElidedNativeAuqIdentity(screen: string, call: NativePlanQuestionCall):
{indices:number[]; packetBar:boolean} | undefined {
const visible = screen.replace(/\r+\n?/g, '\n');
const cursor = [...visible.matchAll(/^❯\s*1\./gm)].at(-1);
if (!cursor) return undefined;
// Native panes can leave blank outer margin rows when the header scrolls
// away. Remove only that margin; internal/unboxed rows remain evidence.
let before = visible.slice(0,cursor.index).replace(/^(?:[ \t]*\n)+/,'').trimEnd();
const bar = call.questions.length > 1
? /^←[^\n]*[☐☒][^\n]*✔[ \t]*Submit[ \t]*→[ \t]*\n/.exec(before) : null;
if (bar) before = before.slice(bar[0].length).replace(/^(?:[ \t]*\n)+/,'');
const rows = before.split('\n');
// Only the actual viewport's boxed question body qualifies. Do not strip a
// foreign header, quoted output, or prose prefix to manufacture a match.
if (rows.length < 2 || !/^[ \t]*[│┃](?: |$)/.test(rows[0]!) || /[☐□❯]/.test(before)) return undefined;
// Once the clipped boxed body starts, every interior row belongs to it.
// Counting may normalize blank rows; capture cannot discard that contradiction.
if (rows.some(row => !/^[ \t]*[│┃](?: |$)/.test(row))) return {indices:[],packetBar:!!bar};
const body = rows.map(row => row.replace(/^[ \t]*[│┃] ?/,'')).join('\n').trimEnd();
if (!body.endsWith('…')) return undefined;
const compact = (value: string) => value.replace(/\s+/g,'');
const fragment = compact(body.slice(0,-1));
if (fragment.length < 160) return undefined;
const menu = visible.slice(cursor.index);
const footer = call.questions.length > 1
? /(?:^|\n)Enter\s*to\s*select\s*·\s*Tab\/Arrow\s*keys\s*to\s*navigate\s*·\s*Esc\s*to\s*cancel[\s│┃─━└┘]*$/.exec(menu)
: /(?:^|\n)Enter\s*to\s*select\s*·\s*↑\/↓\s*to\s*navigate\s*·\s*(?:n\s*to\s*add\s*notes\s*·\s*)?Esc\s*to\s*cancel[\s│┃─━└┘]*$/.exec(menu);
// Once this boxed/elided shape is recognized, an incompatible
// footer is contradictory evidence, not permission to try another route.
if (!footer) return {indices:[],packetBar:!!bar};
// Inspect every displayed row, including an out-of-order control that the
// shared parser intentionally omits when its contiguous menu ends.
const options = [...menu.matchAll(/^[ \t]{0,3}(?:❯[ \t]*)?([0-9]+)\.[ \t]*(.*)$/gm)]
.map(match => ({index:Number(match[1]),label:match[2]!}));
const matches = call.questions.flatMap((question,questionIndex) => {
const offered = options.slice(0,question.options.length);
const controls = options.slice(question.options.length);
const exactMenu = offered.length === question.options.length && offered.every((option,index) =>
option.index === index + 1 && compact(option.label) === compact(question.options[index]!.label)) &&
controls.length <= 2 && controls.every((option,index) =>
option.index === question.options.length + index + 1 &&
(index === 0 ? /^Typesomething\.?$/ : /^Chataboutthis$/).test(compact(option.label)));
if (!exactMenu) return [];
const native = compact(question.question);
// A literal terminal ellipsis can also be complete-body/suffix evidence.
// Count that candidate alongside elided candidates, before either route
// can select a different member of the same packet.
if (bar ? native === compact(body) : native.endsWith(compact(body))) return [questionIndex];
const start = native.indexOf(fragment);
if (start < 0 || start + fragment.length >= native.length) return [];
// Repeated visible segments remain ambiguous, not an absent candidate
// that would let another packet member win by default.
return native.indexOf(fragment,start + 1) === -1 ? [questionIndex] : [questionIndex,questionIndex];
});
return {indices:matches,packetBar:!!bar};
}
/** Project a verified complete body; null rejects contradictory boxed evidence. */
function unboxCompleteNativeAuqBody(screen: string, call: NativePlanQuestionCall): string | null | undefined {
// Packets already have their own body projection and unique-tab matching.
// A singleton header must never provide a new route into a packet.
if (call.questions.length !== 1) return undefined;
const visible=screen.replace(/\r+\n?/g,'\n');
const cursor=[...visible.matchAll(/^❯\s*1\./gm)].at(-1);
if (!cursor) return undefined;
const before=visible.slice(0,cursor.index);
const header=[...before.matchAll(/(?:^|\n)[ \t]*[☐□]([^\n│]*)\n/g)].at(-1);
if (!header) return undefined;
const compact=(value:string)=>value.replace(/\s+/g,'');
const question=call.questions[0]!;
const bodyStart=header.index+header[0].length;
const bodySpan=before.slice(bodyStart);
const rows=bodySpan.replace(/^(?:[ \t]*\n)+/,'').trimEnd().split('\n');
if (!/^[ \t]*[│┃](?: |$)/.test(rows[0]!)) return undefined;
if (compact(header[1]!)!==compact(question.header) || rows.some(row=>!/^[ \t]*[│┃](?: |$)/.test(row))) return null;
// A complete boxed pane must pass this capture-specific check before the
// broader counting parser. Only a genuinely elided body uses its other route.
const body=rows.map(row=>row.replace(/^[ \t]*[│┃] ?/,'')).join('\n');
if (!compact(body) || compact(body)!==compact(question.question)) return body.endsWith('…') ? undefined : null;
// Ordinary prior public output is outside the pane only when separated by
// the native horizontal rule. A quoted/fenced pane is not display evidence.
const prefix=before.slice(0,header.index).trimEnd();
if (prefix && (!/^[─━]{20,}$/.test(prefix.split('\n').at(-1)!) ||
/(?:^|\n)[ \t]*(?:>|```)/.test(prefix))) return null;
// Strip exactly one framing prefix, preserving literal box characters in
// the actual question. No size threshold: malformed short briefs need grades.
const menu=visible.slice(cursor.index);
if (!/(?:^|\n)Enter\s*to\s*select\s*·\s*↑\/↓\s*to\s*navigate\s*·\s*(?:n\s*to\s*add\s*notes\s*·\s*)?Esc\s*to\s*cancel[\s│┃─━└┘]*$/.test(menu)) return null;
const options=[...menu.matchAll(/^[ \t]{0,3}(?:❯[ \t]*)?([0-9]+)\.[ \t]*(.*)$/gm)]
.map(match=>({index:Number(match[1]),label:match[2]!}));
const offered=options.slice(0,question.options.length), controls=options.slice(question.options.length);
if (offered.length!==question.options.length || offered.some((option,index)=>option.index!==index+1 || !option.label.trim()) ||
controls.length>2 || controls.some((option,index)=>option.index!==question.options.length+index+1 ||
!(index===0?/^Typesomething\.?$/:/^Chataboutthis$/).test(compact(option.label))) ||
!nativeOptionLabelsAgree(visible,question)) return null;
return visible.slice(0,bodyStart)+bodySpan.replace(/^[ \t]*[│┃] ?/gm,'')+visible.slice(cursor.index);
}
/** A hook payload alone, screen prose, or a different packet is not a capture. */
export function displayedNativeAuq(screen: string, call: NativePlanQuestionCall | undefined):
{question: NativePlanQuestion; questionIndex: number} | undefined {
if (!call || call.answered || call.failed) return undefined;
const clipped = clippedElidedNativeAuqIdentity(screen,call);
const uniqueFirst = clipped?.indices.length === 1 && clipped.indices[0] === 0;
// Question identity must be unique across complete and elided body routes;
// an early shared match must not bypass a contradictory second candidate.
if (clipped && !uniqueFirst) return undefined;
const unboxed=unboxCompleteNativeAuqBody(screen,call);
if (unboxed===null) return undefined;
const matched = capturePlanCountQuestion(unboxed ?? screen, new Set(), 0, false, call);
// Never use the screen-only fallback. The additional native-only branch
// proves the observed combined clipping mode from this same owned payload.
// A visible tab bar keeps its existing route; its parsed body above is only
// additional ambiguity rejection, never a new route to accept another tab.
if (!matched?.nativeCall && (!uniqueFirst || clipped?.packetBar)) return undefined;
const questionIndex = matched?.nativeCall ? matched.nativeQuestionIndex ?? 0 : 0;
if (questionIndex !== 0) return undefined; // A later tab cannot replace the first question.
const question = call.questions[questionIndex];
return question && nativeOptionLabelsAgree(screen, question) ? {question, questionIndex} : undefined;
}
/** Read only a public CLI error panel; never inspect private journal blocks. */
export function nativeAuqPublicError(screen: string): string | undefined {
const match = new RegExp(`${API_ERROR_PANEL.source}[\\s\\S]*`, 'i').exec(screen);
return match?.[0].trim().slice(0, 2000);
}
/** The current viewport only, never terminal history or a model transcript. */
export function nativeAuqViewport(screen: string): {viewport:string; viewportTruncated:boolean} {
const limit = 16_384;
return {viewport:screen.slice(-limit), viewportTruncated:screen.length > limit};
}
export async function captureNativeFirstAuq(opts: NativeAuqCaptureOptions): Promise<NativeAuqCapture> {
const startedAt = Date.now();
const deadline = startedAt + NATIVE_AUQ_CAPTURE_MS - CLEANUP_MS;
const cwd = path.resolve(opts.planDir);
const model = resolveEvalModel('capture', opts.model);
const sessionId = randomUUID();
const runId = opts.runId || process.env.EVALS_RUN_ID || `local-${sessionId}`;
let session: ClaudePtySession | undefined;
let ownedRoot: string | undefined;
let artifactDir: string | undefined;
let artifactRoot: string | undefined;
let outcome = 'error';
let diagnostic: string | undefined;
let captured: NativeAuqCapture | undefined;
let pendingPublicCall: NativePlanQuestionCall | undefined;
let viewport: ReturnType<typeof nativeAuqViewport> | undefined;
let viewportAt: string | undefined;
let displayMatched = false;
let pendingRecorder: ReturnType<typeof pendingQuestionRecorderStatus> | undefined;
const observeViewport = async (): Promise<string> => {
const screen = await session!.currentScreen();
viewport = nativeAuqViewport(screen);
viewportAt = new Date().toISOString();
return screen;
};
const fail = (reason: string, detail?: string): never => {
outcome = reason;
diagnostic = detail;
throw new Error(`${opts.testName}: AUQ capture failed (${reason})${detail ? `: ${detail}` : ''}`);
};
try {
if (!isHermeticEnabled()) fail('hermetic_required');
if (process.env.GSTACK_EVAL_DIR || process.env.EVALS_RUN_ID || opts.runId) {
const segment = (value: string) => value.replace(/[^a-zA-Z0-9_-]/g, '_').slice(0, 120) || 'run';
// Resolve the optional store inside the operation budget, before launch.
artifactRoot = path.resolve(process.env.GSTACK_EVAL_DIR || getProjectEvalDir(), 'native-auq', segment(runId));
}
// This invocation owns its session/config. Only this fresh recorder's exact
// public PreToolUse fields are read; its parent JSONL is never opened.
ownedRoot = fs.mkdtempSync(path.join(os.tmpdir(), 'gstack-native-auq-'));
const configDir = path.join(ownedRoot, '.claude');
const stateDir = path.join(ownedRoot, 'gstack-home');
fs.mkdirSync(configDir);
fs.mkdirSync(stateDir);
fs.writeFileSync(path.join(configDir, '.claude.json'), JSON.stringify(buildSeedConfig({
apiKey: process.env.ANTHROPIC_API_KEY ?? process.env.GSTACK_ANTHROPIC_API_KEY,
trustedDirs: [cwd],
})), {mode:0o600});
const skillPath = path.join(cwd, opts.skillName, 'SKILL.md');
const prompt = `The ONLY skill file you may read is this absolute path: ${skillPath}, together with the files in its sections/ directory that it tells you to Read. Do NOT search for, Glob, find, or read any other SKILL.md anywhere — especially nothing under ~/.claude or /Users.
Read ${skillPath} and follow its workflow for this scenario:
${opts.scenario}
Skip any system-audit / environment-setup / codebase-exploration steps. At the first decision requiring user input, ask the user through the AskUserQuestion tool and wait for their answer.`;
session = await launchClaudePty({
// Explicit tool approvals suffice. Bypass mode introduces a separate
// safety-consent screen in a fresh CLI config; never accept that screen.
cwd, model, seedSkills: false, permissionMode: 'default',
observeScreen: true, observeSetupQuestions: true,
timeoutMs: Math.max(1, deadline - Date.now()),
// Explicit availability as well as approval: no Bash/Agent/Skill or MCP.
// The positional prompt starts a real interactive turn, without a boot
// sleep, hypothetical output request, or synthetic answer submission.
extraArgs: ['--tools', 'Read,Write,AskUserQuestion', '--allowed-tools', 'Read,Write,AskUserQuestion',
'--session-id', sessionId, prompt],
env: {CLAUDE_CONFIG_DIR: configDir, GSTACK_HOME: stateDir, GSTACK_HEADLESS: ''},
});
if (session.hermeticConfigDir !== configDir || !session.pendingQuestionFile) fail('missing_observer');
while (Date.now() < deadline) {
const screen = await observeViewport();
const call = readFirstPendingQuestionForDisplay(session.pendingQuestionFile, cwd, configDir, startedAt, sessionId);
if (call) pendingPublicCall = call;
pendingRecorder = pendingQuestionRecorderStatus(session.pendingQuestionFile, cwd, configDir);
const publicError = nativeAuqPublicError(screen);
if (publicError) fail('error_api', publicError);
if (session.exited()) fail(`exit_code_${session.exitCode() ?? 'unknown'}`);
if (pendingRecorder.status === 'invalid') fail('invalid_capture', pendingRecorder.reason);
const displayed = displayedNativeAuq(screen, call);
if (displayed && call) {
captured = {outcome:'question_captured', workflowCompleted:false, source:'pre_tool_use',
sessionId, toolUseId:call.toolUseId, ...displayed, publicCall:call,
text:serializeNativeAuq(displayed.question)};
displayMatched = true;
outcome = captured.outcome;
break;
}
await Bun.sleep(Math.min(50, Math.max(0, deadline - Date.now())));
}
if (!captured) {
// Refresh the final public frame before shutdown; a startup/permission
// screen is useful timeout evidence even when no AUQ hook ever fired.
const screen = await observeViewport();
const call = readFirstPendingQuestionForDisplay(session.pendingQuestionFile, cwd, configDir, startedAt, sessionId);
if (call) pendingPublicCall = call;
pendingRecorder = pendingQuestionRecorderStatus(session.pendingQuestionFile, cwd, configDir);
const publicError = nativeAuqPublicError(screen);
if (publicError) fail('error_api', publicError);
fail('timeout');
}
} catch (error) {
if (outcome === 'error') diagnostic = String(error);
throw error;
} finally {
// Deliberate cutoff after a displayed question is not workflow completion.
let cleanupError: string | undefined;
let artifactError: string | undefined;
try { await session?.close(); } catch (error) { cleanupError = String(error); }
try { if (ownedRoot) fs.rmSync(ownedRoot, {recursive:true, force:true}); }
catch (error) { cleanupError = [cleanupError, String(error)].filter(Boolean).join('; '); }
if (captured && cleanupError) outcome = 'cleanup_error';
// Persist only supported public fields. No terminal history or transcript
// is retained, and no partial artifact can supply a later invocation.
try {
if (artifactRoot) {
const segment = (value: string) => value.replace(/[^a-zA-Z0-9_-]/g, '_').slice(0, 120) || 'run';
fs.mkdirSync(artifactRoot, {recursive:true, mode:0o700});
artifactDir = fs.mkdtempSync(path.join(artifactRoot, `${segment(opts.testName)}-`));
fs.writeFileSync(path.join(artifactDir, 'capture.json'), JSON.stringify({
...captured, pendingPublicCall, ...viewport, viewportAt, displayMatched, pendingRecorder,
outcome, workflowCompleted:false, diagnostic, cleanupError, testName:opts.testName,
skillName:opts.skillName, model, runId, sessionId, cwd,
// Native display capture does not expose a terminal billing result.
billing:'unavailable',
elapsedMs:Date.now() - startedAt, at:new Date().toISOString(),
}, null, 2) + '\n', {mode:0o600});
}
} catch (error) { artifactError = String(error); }
if (captured && artifactError) outcome = 'artifact_error';
console.log(`[AUQ-native ${opts.testName}] outcome=${outcome} workflowCompleted=false`
+ (artifactDir ? ` artifact=${artifactDir}` : '')
+ (artifactError ? ` artifactError=${artifactError}` : '')
+ (cleanupError ? ` cleanupError=${cleanupError}` : ''));
// Keep the original refusal/crash/timeout primary when diagnostics fail.
if (captured && (cleanupError || artifactError)) {
throw new Error(`${opts.testName}: AUQ capture failed (${outcome}): ${cleanupError || artifactError}`);
}
}
return {...captured!, artifactDir};
}