Garry Tan
55f8151fbe
* docs: approved reliability follow-ups plan (Oct 2026)
* D1: score overlay wrong answers as comparison results (contract v4)
A completed answer that fails the fixture's output check is now a valid
measurement with taskCorrect false and answerError set, in either arm.
Only execution, scope and metric failures invalidate a trial. ON
correctness still gates the case, so an OFF wrong answer stays in the
comparison. OVERLAY_CONTRACT moves to v4; v3 verdicts are not
reinterpreted. Replays the 37198445662 sonnet trials (OFF-0 wrong).
* C4: score auq-matrix substance with a 3-sample judge panel
Each skill keeps one native capture in the same execution; the
recommendation-substance grader now runs as a JUDGE_PANEL_SAMPLES panel on
that capture and gates on the mean against the unchanged minimum of 4. A
failed sample fails the panel without resampling. Approved at the gate
2026-10-04 for censuses after this change.
* E1: one Bun version source; setup refuses below the 1.3.3 floor, updater checks before advancing
bin/gstack-bun-version.sh holds the security floor (1.3.3) and the tested
version (1.4.0, the CI pin and engines.bun). setup refuses Bun below the
floor before writing anything, warns between floor and tested, warns and
continues on an unparseable version, and prints a 1.4.0 install hint.
Prereleases sort below their release (semver).
gstack-session-update now fetches, reads the incoming release's floor and
checks the bun setup will run before fast-forwarding to that exact
revision; below it the update is held with reason bun-too-old and a
resume command, leaving the revision and installs untouched.
Fixture copies of setup (team-mode, claude-code migration) now include the
sourced file; the slow-git lock test sleeps on fetch, the updater's
network step.
* E3: record the owned Chromium's PID over CDP so the #2709 reap runs on Playwright 1.62
Playwright 1.62's Browser has no process(), so chromiumProcInfo was always
null and browse stop never reaped a surviving headless Chromium. The owned
headless launch now reads the browser PID from CDP SystemInfo.getProcessInfo
(bounded at 2s; a hung or failed call logs once and records nothing). The
reaper's start-time and cmdline identity checks are unchanged, so a recycled
PID or an externally owned browser is never signalled.
* A1: ship docs gate records post-child hashes with gstack-docs-candidate, saves each record once
Stored transcripts (37198445662 gate-census-2) show the parent's reads of
release files take ~1 s; the parent's own model time dominates (completion
268 s of 392 s, store ~270 s of 387 s), and 45-78 s of it is retyping hash
maps, child output and records into several artifacts after the child
returns. The child contract is unchanged.
* A2: /plan-ceo-review sends the mode handoff line before any other tool call
plan-ceo-mode-routing's HOLD SCOPE session spends its 240 s post-selection
window on model work: 37176837432 skipped the handoff helper and paraphrased
the mode, 37198445662 ran it, then wrote the plan, loaded sections and probed
Codex for 114 s before the posture chat. The handoff line is the posture.
* B3: accept wrapped unchanged rows in the cropped Edit preview
PR run 37176835584 timed out with the floor's owned Edit menu on screen:
unchanged diff rows soft-wrap under their gutter with a blank marker
column, and a trimmed empty unchanged row keeps only its number. The
parser accepted wrapped +/- rows only, so it never verified the owned
edit. Both forms are now parsed and still compared with the exact source
lines; the captured viewport is replayed with negative controls.
* A3: record per-phase headroom diagnosis for docsync, plan-mode-no-op, mode-routing and gate slice 11
* E4: /review Step 1 keeps going when the base fetch fails and reports stale base coverage
Under Codex's read-only .git, 'git fetch && merge-base && diff --stat'
stopped at the fetch, so the empty output read as 'Nothing to review'.
The fetch is now separate, prints BASE_REFRESH fresh or stale with the
local base revision, and the review states stale coverage.
* D2: reword the format-approach question prompt
Both plan-ceo-review-format-approach refusals in census 37195203538
(Fable 5.1 model_refusal_no_fallback, category reasoning_extraction:
"duplicating model outputs") were anchored on the session prompt, whose
capture instruction asked for the "verbatim" question and "the exact
prose you would have shown" in a "format-capture test". The approach
case now asks for the same question in the same file as the step's
output, without that wording. The other three plan-format cases keep
their unchanged prompt (no refusals in 72 trials).
* C2: /qa defines its outcome filename fields; QA report templates name the measured timing fields
Low judge samples in failing panels named both: qa workflow (37186854666)
'an undefined {user} in the outcome filename', and qa-only workflow
(37198445662) 'template asks for Duration / durationMs totals while
reporting.md mandates Probe budget and Guarded command time'. The qa
union stays under its 1.11 size ratio (93292 of 93294 bytes).
* 0.1: session ledger written by the claude -p, Agent SDK, PTY and Codex runners
Each armed model session appends one best-effort line to
$GSTACK_EVAL_DIR/session-ledger.jsonl: a stable session key, the timeout it
armed, monotonic elapsed time, its end reason (completed, session_timeout,
observer_timeout, api_error, refusal, aborted, error), one sanitized evidence
line and, for stream-JSON runners, a compact liveness summary (longest
in-request silence with nothing outstanding, last event kind, tool open at
end). Partial messages now stream for every claude -p session and for SDK
sessions without a custom query provider (runSharedInteractive opts in); their
rows feed only the summary, so transcripts keep their size unless
publicStreamDiagnostics is set. Structured refusal and API-error signals win
over phrases, and tool results are never read for them.
* 0.2: failure_cause beside failure_class, from runner facts and the session ledger
One pure failureCauseOf() beside failureClassOf() sets failure_cause
(contract > pre_turn_infra > api_error > refusal > provider_stall >
session_timeout > observer_timeout > assertion > unknown; no machine-set
'detector') with one bounded evidence line, and failureDetailOf() keeps a
Bun matcher's Expected/Received or a judge's failing dimensions in
failure_detail. Both record builders (classifyTrialShard, junitCensus) call
them and attach the trial's ledger sessions. failure_class, error and
TRIAL_ERROR_MAX are computed exactly as before; the new fields are optional,
bounded and validated only when present, and a frozen copy of the v1 reader
accepts every new record. A provider stall needs partial messages and
STALL_WINDOW_MS (120 s, eval-budgets.ts beside HEADROOM_WARN/FAIL) of
in-request silence with nothing outstanding; it stays a failure under
EVAL_POLICY v1.
* E4: fingerprint under a read-only .git via a private object dir owned by review-log
Codex's sandbox mounts .git read-only, so gstack-wtree's git add -A could not
write blobs and /review lost its start capture. gstack-review-log now creates
a private 0700 object directory under TMPDIR, passes it to gstack-wtree (which
writes new objects there with the common-dir object store, resolved before
exporting, as an alternate, so linked worktrees work), exposes it to the
bindReview/checkSharedLibsReuse consumers that read the tree and blobs, and
removes it on EXIT/INT/TERM after they finish. gstack-wtree prints one stderr
line and exits 1 on a failed write or unwritable TMPDIR instead of printing a
partial hash; review-log no longer swallows that line. Standalone gstack-wtree
keeps writing to the repository store so git diff between saved snapshots
(/ship) still works.
* 0.3: red lines name their values, evidence and an after-a-repair command; one published-text sanitizer
Each red line in the census report keeps its stored error and adds the
machine cause with its evidence, the failure_detail values (Expected/Received,
or each failing judge dimension with its mean, sample count and the sample
sentence naming it), the slice artifact and shard directory that hold the
transcript, a fetch command (eval:pass-rates --run <id> --case <id>), and
'after a repair:' with a command that selects only that case (E2E ids via
--case, judges via their selection and an exact test-name filter). A
session that completed and failed a check says so instead of 'assertion
(success)'. Report text that reaches a PR comment, issue or committed file
goes through scripts/lib/published-text.ts: control and bidi characters
stripped, credential-scan patterns redacted, mentions neutralized, HTML and
link brackets escaped inline, blocks in a fence longer than any backtick run,
length capped. Verdicts and collector verdict objects are unchanged.
* E2: one make-pdf resolver for readiness and every $P block on env-var hosts
make-pdf exported $P from its readiness block, which Codex and the other
env-var hosts lose between blocks. binaryAssignment gains 'make-pdf': the
MAKE-PDF SETUP readiness check and the runtime prelude both use it, so every
block that runs $P derives the same make-pdf/dist/pdf binary, with
MAKE_PDF_BIN (dev builds) winning when it is executable. Claude output keeps
its repo-local-first probe; only the false 'subsequent blocks' comment
changed. Root plus P=... fits the 400-byte prelude budget; the fixed
MAKE_PDF_OVERRIDE segment is the one exemption, named in runtime-root.ts and
the conformance test.
* 0.1: eval:pass-rates --headroom, --reds and --run report views; 85% headroom alarm in --gate
A new pure module, scripts/lib/eval-history.ts, reports per case the slowest
armed session against the timeout it armed (censored when it timed out,
insufficient below 3 samples, unknown without a session ledger, with the
case wall shown separately as an upper bound), the census critical path from
the slice jobs' timings, the verdict red ledger per census by failure class
and machine cause with shrunk per-case rates, and the all-green probability
(an approximation: the product of 1 - p_i, with an interval). --run <id>
lists one census's reds with their values and history and downloads only the
slice artifacts those reds name, reporting any over 64 MB instead of skipping
them. eval-flake-rank.ts only routes flags: --help works offline, an unknown
flag or case id exits 2 before any fetch, --case narrows the JSON and the
alarms, every view prints its run/branch scope, and --gate also alarms on a
session above HEADROOM_FAIL (85%) of its budget.
* E5: GSTACK_DESIGN_IMAGE_MODEL override; the live model check always tests the defaults
The image_generation tool model gains an env override validated against a
conservative gpt-image-<version> pattern; a bad value fails before any
request and the error names the variable. Pairing and rejection hints name
the tool model actually requested. design/scripts/live-model-check.ts now
ignores GSTACK_DESIGN_MODEL and GSTACK_DESIGN_IMAGE_MODEL (and says so), so
an override cannot mask a retirement of a default, and its exit 2 (no key)
says nothing was checked.
* C3: /deslop-shared-libs names a concrete risk for each recommendation, never 'none'
Both shared-libs-opportunity-judgment reds (37193478719 t3, 37195203538 t1)
failed adoption_and_risk because the top recommendation said 'Main risk:
none found inside the repo' / 'none in behavior'. The skill asked to 'name
the main risk or uncertainty' without saying what counts.
* B4: --case selects loop-registered cases; an unknown id exits before any process
caseSelection() resolves the one paid file that registers a case and how a
trial selects it: by its literal Bun test name, by the id a registration loop
names its test at runtime (coverage audits, TPA, journeys), or by running the
whole file when it registers no other case (plan-mode-no-op, whose test names
are not the id, so the old name pattern ran nothing). A whole-file case runs
one file shard per trial and is judged by shard status, as CI judges it.
--list prints the selection. Every E2E_TIERS id now resolves except
carve-section-loading, which fifteen files register (CI runs each as its own
shard) and which still fails with the owning files named.
* 0.3: after-a-repair runs the shard file when --case cannot resolve a case
A case that several files register (carve-section-loading) cannot be selected
by --case; its red line now prints the bun test command for the shard file
that went red, so every printed command selects exactly one case.
* C1: pre-register workflow-judge control bundles for qa, qa-only, review and setup-deploy
Positive (production bundle), negative (headings only) and one targeted
execution-blocking defect per judge, one 3-sample panel each, built with the
eval's own reader, prompt builder, options and floors.
* F: cost_known: a trial whose harness captured no billing says so; totals never hide it
cost_usd stays a finite number. Trial and eval records gain an optional
cost_known boolean (absent means known, so old records still parse; the
validator accepts only a boolean). Session-ledger rows record whether the
session's terminal result carried total_cost_usd; PTY and Codex sessions
never do, and SDK captures without a terminal result do not. A trial is
cost-known only when it has eval records and every record and session
captured billing, so JUnit-only PTY cases (125 of 298 trials in run
37198445662 recorded 0) now read cost unknown. The census headline prints the
known sum plus the unknown count, never a bare total over unknown costs, and
no total when every cost is unknown; the history import keeps the flag.
PTY transcripts are not priced: test/helpers/pricing.ts estimates exclude
cache writes, so they cannot give the billed cost.
* B1: replay corpora for the five repeat-offender cases
shared-libs-plan-callers, plan-ceo-section-loading, auto-decide-preserved,
plan-design-review-plan-mode and ship-docsync-late-result each get an
inventory of every wave census in scope (all artifacts still available)
and minimal, redacted detector or actor inputs replayed through the real
detector in the free suite: census passes, every distinct census red, and
derived negative controls labelled as derived. Fixed false reds stay as
expected passes with their history; the two genuine auto-decide product
failures (37176837432, 37179171083) stay refused. A contract test bounds
each entry at 64 KB, scans it for HIGH credentials, requires a pass and a
refusal per case, and reports entries whose series identity is stale.
Building the auto-decide corpus exposed a latent native detector misread:
a bold 'Mode application and rationale.' heading was parsed as a
withdrawn Mode field (37186854666). A field status now ends at its
sentence.
* B2: read approved stale-fill decisions from more valid ledger records
The structured ledger reader is authoritative before the prose fallback,
but it missed two valid approved records in passing censuses
(37186854666: a camelCase writeProfile subject; 37195203538: a
slash-joined 'D1 / STALE-FILL' row id), which then rested on the prose
detector that caused both historical reds. Both forms are accepted; an
unresolved row, an unrelated decision, the copied fixture plan and
non-write camelCase words stay refused.
* C1: record workflow-judge control results (negatives separated; qa and review miss a removed step)
* C2: record every low workflow-judge sample in the wave censuses with its disposition
* 0.3: published lines cannot break a fixed fence
evals.yml wraps collector-outcomes failures in a fixed three-backtick fence
for the PR comment, so those lines (and eval:pass-rates output meant for the
tracking issue) go through sanitizeFixedFenceLine: cleaned like every
published line, with backticks turned into quotes. Inline published text
also neutralizes backticks.
* A1: fit the post-child record wording under ship's 1.428 parity size cap
* E4: inline the Step 1 base refresh to fit review's 1.199 parity size cap
* B2: inventory of gate and periodic phrase-only detector checks
docs/evals/detector-inventory.md lists every pass/fail decision in gate
and periodic E2E cases that rests on phrases in model free text, with
its structured source (if any), whether the contract is user-visible,
and a disposition: already structured-first, convert candidate, phrase
witness required, or no structured source. Convert candidates are
ordered by census red history.
* 0.2: failureCauseOf takes the stall window instead of importing it; the PTY ledger never throws
eval-store.ts imported STALL_WINDOW_MS from eval-budgets.ts, which free tests
replace with mock.module stubs (review-finalization-budget, office-hours,
qa-fix-loop, qa-only), so loading eval-store failed there. failureCauseOf is
now dependency-free: the record builders pass STALL_WINDOW_MS. A PTY session
whose handle cannot show its text (test doubles) records no panel instead of
throwing from runPtySession's cleanup. The 37186854666 replay also runs
through classifyTrialShard.
* A2/auto-decide: the CEO handoff chat repeats the helper's printed line verbatim
Claude Code 2.1.284 collapses the helper's Bash output ('Ran 1 shell
command'), so the 'Auto-decided review mode → …' line reaches the user only
through chat. In lane D's auto-decide-preserved corpus every passing capture's
chat starts with that line, and the reds paraphrased it ('HOLD SCOPE was
auto-selected…', 'Mode decision done…', "I'm defaulting to HOLD SCOPE…").
The template now says verbatim and why.
* C2: record the judge panels rerun on the final ship and review wording
* integrate: touchfiles and module-size ratchet entries for the wave's new modules
GLOBAL_TOUCHFILES gains session-ledger.ts, published-text.ts and
auq-substance-panel.ts; bin/gstack-bun-version.sh joins every entry that
lists setup or the updater; auq-matrix stays a rule case with its C4
evidence beside E2E_KINDS; eval-history, published-text and session-ledger
are registered new modules.
* E1: BROWSE SETUP install hint pins the tested Bun (1.4.0)
* E2: quote "$P" in make-pdf commands so install paths with spaces work
* E1: /gstack-upgrade checks the incoming Bun floor before pulling
gstack_bun_incoming_hold (bin/gstack-bun-version.sh) is now the one
pre-advance check: the auto-updater calls it before its fast-forward and
/gstack-upgrade's git path calls it after fetching. Below the incoming
release's floor the upgrade stops with BUN_TOO_OLD and the checkout,
VERSION and installs stay untouched. Vendored installs were already safe:
setup refuses before writing and the swap restores the previous install.
* 0.3: only main posts or closes tracking issues; reports carry alarms, red ledger, headroom and all-green
A branch-dispatched periodic or marathon run writes its report to the step
summary and a report artifact; the upsert and close steps run only on
refs/heads/main. The periodic body now keeps every pass-rate alarm line (the
old grep kept only the header), adds the red ledger and session headroom read
with an explicit --branch, the all-green estimate and the triage guide link,
and routes report text through scripts/lib/published-text.ts. The marathon
lane gains the matching close-on-green step.
Migrated: evals-workflow-wiring pinned report-summary.md in the upsert step;
the body is now written by 'Write the census report'.
* C3: prosons neutral-posture detector ignores a negated "taste call"
Both plan-review-prosons-neutral-neg reds (37179171083 t3, 37193478719 t3)
recommended A and closed with "a coverage call, not a taste call"; the
old /taste call/i matched the negation. The detector moves to a helper so
the two census captures replay free as negatives, with positive controls.
* B2: credit the CEO mode handoff only when the chat shows its line
Claude Code 2.1.284 collapses the gstack-ceo-mode-handoff Bash result, so
the printed AUTO_DECIDE line is user-visible only when a later chat message
of the session begins with it. The native detector now requires that
witness; the helper's tool result locates it but no longer replaces it, and
a generic mode statement cannot stand in once a handoff ran. Census
37182865432 is relabelled a genuine miss (expected: refused): the chat said
"I'm defaulting to HOLD SCOPE…" and "Mode decided: HOLD SCOPE". Corpus
replay: 5 credited, 3 refused. ceo-mode-handoff tests migrated to the
witness rule (deliberate behavior change).
* test: baseline the wave's approved detector and harness replay tests
B1 corpora, B3, C3, C4 and D2 replay captures through paid-case detectors,
actors or fixtures (test code by design); each entry carries its reason.
* E6: live Codex multi-block periodic case in the host-run Codex job
test/codex-e2e-multiblock-live.test.ts runs ./setup --host codex once from the
job's checkout into a fresh HOME and CODEX_HOME (auth.json copied from the CI
login home), asserts codex debug prompt-input lists the gstack router as a
real file, seeds one learning, and runs /learn's Show recent block through
codex exec: that later block must root itself through its prelude, run the
installed gstack-learnings-search and print the seeded learning; no block may
report a missing install or an unresolved gstack helper path.
The file joins CODEX_CI_FILES so the planner gives it to the host-run Codex
job. paid-test-durations.json carries a 180 s packing estimate (refreshed by
--write-durations after the first census) so the plan keeps every slice
within max-parallel. Oracles live in test/helpers/codex-multiblock.ts with
free replays in test/codex-multiblock.test.ts. Registration (E2E_TOUCHFILES,
E2E_TIERS periodic, E2E_KINDS rule) is the integrator's.
* E5: design live-model check as a periodic case
test/skill-e2e-design-model-smoke.test.ts runs design/scripts/live-model-check.ts
(default models only) under a temp HOME. Without OPENAI_API_KEY the case is a
Bun skip that the census lists as SKIPPED with no coverage credit; with a key,
exit 0 passes and any other exit fails with the script output (exit 2 under a
present key fails as 'nothing was checked'). paid-test-durations.json carries
a 90 s packing estimate. Registration (E2E_TOUCHFILES, E2E_TIERS periodic,
E2E_KINDS rule) is the integrator's.
* A1: docsync parent points children at a saved interface file
The /ship docsync fixture saved its observation interface only in the
parent prompt and told the parent to copy it into every child prompt,
about 30-40 s of parent generation per session (lane S, 37198445662).
docsSessionOptions now writes it to <home>/fixture-interface.md (0600) and
the parent tells each child to Read that file first. Fault-actor prompts,
which prepare already writes to a file, are unchanged. Fixture-only.
* A3: a JUnit case's recorded wall is capped by its shard's wall
junitCensus summed a case's JUnit test times as duration_ms; for
plan-mode-no-op's concurrent PTY tests that is 346 s inside a 176 s shard
(the "432-613 s" history). The case wall headroom shows is now
min(sum, shard elapsed), a true upper bound. Session headroom already
used the ledger's per-session clocks.
* A3: refresh paid-test durations from census 37198445662
Recorded 2026-09-29 estimates under-packed gate slice 11 (17 files
estimated 1078 s ran 1353 s; plan-review-report 76 s -> 198 s,
plan-ceo-plan-mode 37 s -> 112 s, safety-codex-consult 83 s -> 150 s).
Refreshed both tiers with --report --write-durations from the latest
census's slice walls. Old plan re-costed with observed walls: worst slice
~677 s (slice 11); refreshed plan: every slice estimated <= 539 s.
* C2: Browser Quick gets a feasible budget; QA loop names its directory and stop path
Judge rationales on the qa-only panels (2 of the last 3) cited the 30 s
Browser Quick budget: homepage + 5 targets, each a checkpoint write and a
guarded command, left ~5 s per probe. Quick is now 180 s (method text: 3
minutes); the qa-only-no-fix paid check expects the same native deadline.
PROBE_DIR points at its source line, a stopped loop goes to the final
report steps (annotations, materialize, Markdown), and /review and /ship
say that after smoke expiry only plan checks rerun in revalidation.
* C2: name the flows judges found implicit in review resolvers
Structured outside review: MISSING COVERAGE asks no fix question and is
never reported clean (was "preserve the existing user decision flow").
/review's HIGH-impact plan gate: spawned or non-interactive runs report
REQUIREMENTS MISSING and continue, writing no TODOs. Value cards: the byte
clamp applies to card lines; JSON records you write keep full values.
* C1: a free test catches a removed workflow step, which the judges missed
C1 controls showed the qa workflow judge passing /qa without Phase 10 and
the review judge passing /review without Step 3. The new test pins every
step/phase heading of qa, qa-only, review, ship, plan-ceo-review,
plan-eng-review and setup-deploy (derived from the templates' step
headings) and names any that goes missing or out of order; new step
headings must be pinned. The judge rubric is unchanged (EVAL_POLICY).
* size: fit the wave's wording under ship, review and qa parity caps
Measured after all lane merges: ship -79, review -210, qa -68 bytes over.
Trimmed the added wording first (post-expiry rule, MISSING COVERAGE,
plan-gate default, value-card JSON, stop line) and folded /review's
redundant INFORMATIONAL sentence and findings bullet. ship 268,036 (room
8), review 130,118 (room 1). qa needed 11 more bytes: cap 1.11 -> 1.1102
with the measured 93,305 / 84,049 = 1.11013 and its reason beside it.
* docs: troubleshooting sections for the wave's new messages
Bun floor (setup, auto-update, /gstack-upgrade), /review stale base and
fingerprint temp-dir errors, the design image model override, the
unrecorded Chromium PID, and a contributor section for census report
lines (pass-rates flags, failure causes, headroom, cost unknown, overlay
v4, stale corpora, auq substance). A free test checks every
troubleshooting.md#anchor printed by setup or bin scripts exists.
* TODOS: reliability follow-ups deferrals; close the Bun floor, make-pdf $P and #2709 reap
* 0.4: 'A census went red' guide; docs for causes, session ledger, headroom and new messages
docs/evals/census-red.md walks one red line to a diagnosis: prerequisites, the
report fields, eval:pass-rates --run/--reds/--headroom with expected output,
what unknown/insufficient/unrecorded data means, free replay, and the single
'after a repair' paid run allowed by TESTING_INTERNALS. Linked from
CONTRIBUTING, TESTING_INTERNALS and both census reports.
TESTING_INTERNALS documents the failure_cause glossary (one row per
FAILURE_CAUSES value, diagnostic only under EVAL_POLICY v1), failure_detail,
sessions[], cost_known, the session ledger, the pass-rates views and the 85%
headroom alarm, main-only tracking issues, overlay contract v4, auq-matrix's
3-sample substance panel and the E5/E6 periodic cases. troubleshooting.md
gains #bun-too-old and an anchor per new user-facing message from lanes P, D
and F. README names GSTACK_DESIGN_IMAGE_MODEL. test/census-red-docs.test.ts
pins the glossary to FAILURE_CAUSES, parses every guide command and checks the
links.
* v1.91.20.0: CHANGELOG and version for the reliability follow-ups
* regen: SKILL.md and sections for all hosts, digest, context-budget fixture
bun run gen:skill-docs --host all; goldens and llms.txt re-rendered
unchanged; bun test/helpers/capture-context-budget.ts.
* integrate: register codex-multiblock-live and design-model-smoke (periodic, rule)
* 0.4: census report and eval:pass-rates views link the census-red guide
* test: cookie approval fixture reverses the Bun install-hint pin (E1)
The historical approved cookie-workflow prompt carried BUN_VERSION=1.3.10;
the regenerated install hint now pins 1.4.0, so the fixture's reversal
list gains that edit like the earlier waves' entries.
* test: release floor counts the two new periodic cases (E5, E6)
* fix: /freeze, /guard, /unfreeze and /investigate work on Codex and other env-var hosts
Templates quoted "$HOME/.claude/skills/gstack/...", which the env-var hosts'
rewrites turned into "$HOME/.agents/skills/gstack/..." on Codex (a path a
global install never has; same class on the other hosts), and setup's
runtime roots had no freeze/bin. Quoted runtime paths now use
{{RUNTIME_ROOT}}: $GSTACK_ROOT on env-var hosts (the fence gets the
runtime prelude), the unchanged $HOME path on Claude. _link_runtime_dists
links freeze/bin and the careful/bin helper it sources into every
env-var host's root. Also fixes the same guess in /benchmark-models,
/retro and /diagram. test/env-host-runtime-assets.test.ts fails on the
previous code: no fence guesses a $HOME/<dir>/skills/gstack path, every
$GSTACK_ROOT path a fence uses is in the host's runtime root, and the
installed Codex /freeze and /unfreeze fences run from a fresh shell.
Template-reading tests now read the Claude render or expect the
placeholder (deliberate change).
* CHANGELOG/TODOS: lane W's workflow, cases and guide; the env-var host freeze fix
* E6: a block that reads a SKILL.md is not a missing install
The first live codex-multiblock-live run passed its real check (the /learn
block found the install and printed the seeded learning) but the oracle
flagged a sed of the learn SKILL.md, whose prelude source quotes the
'gstack: no install found (tried $_r)' text. Only an emitted line that
starts with the message and names a real root counts now; free replay
negative added.
* test: qa-only fixture keeps its caller-set 30 s deadline
The qa-only-no-fix fixture tells the model to start the guard with 30 s
(a caller limit), so its deadline assertion's default stays 30 000 ms; the
product's Browser Quick default (180 s) does not apply to it. Reverts the
helper and its test to their base.
* C2: keep PROBE_DIR 'owned' (qa-functional-prompt pins it); qa measured 93,311
* test: paid file count and periodic floor include the two new periodic cases
* test: ship docs freshness pin checks meaning after A1's wording trim
The snapshot still hashes selected release paths (--select narrows them),
generated outputs and docs; the pin now checks that instead of the
pre-trim sentence.
* E3: reap the recorded Chromium only after the daemon has exited
With the PID now recorded (E3), browse stop reaped Chromium while the
daemon was still shutting down; the daemon read that as a crash, exited 1
and left its state file (chromium-profile-isolation's headless stop test
failed 3/3 locally, passes on base). The CLI waits up to 5 s for the daemon
to exit before reaping a survivor.
* regen: qa and qa-only exploratory sections (PROBE_DIR wording)
* B1: batching counter reads a target declared mid-line; replay corpus
GSTA-23's census 37228573062 failed plan-eng-multi-finding-batching with
reviewCount=0 after 975 s although D1-D10 were asked one per finding,
each with its saved decision-ledger record. The report wrote 'Report
destination requested by the user. Review target (fixed): `PLAN.md` ...'
and the counter read target fields only at a line start, so no brief
inherited the plan identity. A target field may now open any sentence of
its line; a mid-sentence mention or another file still gives no identity.
Floor (3), budget and kind unchanged. Corpus: 9 earlier passing censuses,
the red (now credited at D5), and a derived batching refusal; 975 s noted
in the headroom diagnosis.
* B1: shared-libs corpus labels its git path and covers the supported shape
Every census entry was recorded in CI's eval image (git 2.43), where
gstack-safe-git falls back because --no-lazy-fetch is rejected; each entry
now says so (git_path: fallback). Two synthetic supported-path entries
(HEAD and a history citation in the brief, as git >= 2.44 produces) keep
the 37193478719 pass and the 37195203538 refusal, and a test asserts both
shapes are covered and the actor reads no git-path-specific text.
* v1.91.21.0: move off v1.91.20.0 (claimed by #3031)
* test: settle nav-guard commands before asserting; make two new tests Windows-safe
navigation-guard: Bun's expect(promise).rejects slows each Playwright round-trip
to ~1s while it waits (50 page.evaluate: 43ms plain vs 46.6s inside the matcher),
so the click's 5s timeout expired before the guard saw the link-local request
(CI: 'click: Timeout 5000ms exceeded ... waiting for scheduled navigations').
Settle the command first, then assert on the refusal; a source check keeps it so.
env-host-runtime-assets: PROBE_FILES now matches either path separator.
detector-corpus ship-docsync-late-result: JSON-escape the <ROOT> substitution so a
Windows temp path's backslashes do not break JSON.parse.
* fix: CI run/job ids are not phone numbers; context-recovery fixture gets the full runtime bin
pii.phone.e164 redacted the run id in the published eval report's fetch command
(--run 37235771700), breaking the command it prints. Digit-only spans in a run/job
flag, gh run argument, actions/runs/ or /job/ URL segment, or run_id/job_id key are
now exempt; the same digits elsewhere still report.
context-recovery-artifacts: the fixture's ./bin/ lacked gstack-skill-start and
gstack-context-recovery (the preamble's runtime scripts since v1.71), so the agent
spent its 10 turns hunting for them and hit error_max_turns with no final message
(run 37235771700, 11 turns). Link the whole runtime bin; diagnostic rerun passed
in 6 turns.
* test: replay run 37237194905's late-result capture through the docs authority detector
The only failure was call 20's write to a mistyped shard root (3GanyY for 3GanyU),
which the model disclosed; the detector is correct and the A1 post-snapshot
--out under the fixture home stays permitted. Named red recorded in TODOS.
* test: keep docsync-authority in the Windows profile
The replay test's literal helper path tripped the Windows curator's shebang-spawn
pattern, dropping the whole file from the Windows profile (test-free-shards
invariant). The snapshot --out case is already covered by docsync-command-grammar,
so it is removed here, and the Linux-path replay skips on win32.
* fix: gstack-docs-candidate --help works; docs gate red lines name the denying tool failure
PR run 37239016858 ship-docsync-completion: the parent ran
'gstack-docs-candidate --help' (the helper's usage error points at --help, which
failed). That undeclared command made docsToolFailures non-empty, which denied
the atomic-replacement proof of the child's legitimate native Edit, so the red
line blamed the Edit's temp file. A replay of the capture proves the atomic
replace once --help is declared.
- the helper prints its usage for --help/-h and exits 0
- the fixture interface and command grammar admit the helper's bare --help
- the atomic-proof denial names its check and the tool failures behind it
(deny labels renumbered to their lines)
- replay with negative controls (undeclared command, temp left behind, renamed
elsewhere); diagnostic rerun of ship-docsync-completion passed
* test: retro-base-branch fixture uses main and a shallow report path
PR run 37241493377: the agent wrote retro-output.md one directory above the
requested <tmp>/retro-base/retro-output.md (it dropped the retro-base segment),
so the report check found nothing. The fixture also ran plain git init, which
makes master on CI's git while the prompt says the repo is on main; the agent
logged that mismatch as a learning. Init on main and use the temp dir itself as
the repo. Diagnostic rerun passed (18 turns, 203s).
* test: qa-quick sets a caller probe limit and drains past its session budget
C2 raised Browser Quick's native budget to 180 s. qa-quick is the only gate case
on the native budget, and its wall went 194 s (main periodic, pre-C2) -> 254 s and
274 s (this PR's runs) -> 300 s session timeout (run 37242691913): the agent now
probes every navigation target with a checkpoint, then needs ~80 s to report.
Like qa-only-no-fix, the eval now sets a caller probe limit (30 s), scopes the
fixture, and gives bun's outer timeout the session drain margin instead of racing
the 300 s session. The product's 180 s budget and the session budget are unchanged.
Diagnostic rerun passed in 195 s.
* review/ship: use main's post-expiry smoke sentence; v1.91.24.0
GSTA-23's measurement on the review workflow judge: main's 'Post-expiry smoke
rechecks are not-run.' passed 10/10; the C2 wording passed 6/10 and the
'..., even in 4c.' variant failed in PR run 37248650879. qa-probe-gates pins
main's sentence before step 4c.
* test: record the plan-design corpus test in the test-of-test baseline
#3031's ratchet now counts tests that slice a paid test file's source; the B1
plan-design-review-plan-mode replay corpus reads test/skill-e2e-design.test.ts
like its sibling corpora already listed.
* fix: --case no longer deadlocks importing the running CLI
runCaseDiagnosis loaded runPaidShards with a dynamic import of
scripts/test-paid-shards.ts; when that file is the entry it is still in its
top-level await, so the import never resolved and `--case` hung after its
first line (on main since #3031, found by the per-case measurement runs). The
CLI now passes its runner in; library callers keep the lazy import.
* test: name the in-flight file and the native error when bun dies before its summary (#3037)
* Bun 1.4.2 everywhere CI and setup pin the tested version
Bun 1.4.0 on Windows aborts about 3% of free-suite shards with native
bad-handle errors (main too; #3037); the same head on 1.4.2 had none in 180
shard runs. Every current 1.4.0 pin moves together, as bun-version-drift
requires: setup-bun in every workflow (CSO orchestration included), the eval
image (sha256 from bun-v1.4.2's SHASUMS256.txt, re-hashed locally), GitLab,
the Ubicloud free-suite setup, gstack-bun-version.sh's tested version,
engines.bun and @types/bun, the install hint in generated skills, the Dia and
cookie native-qualification pins, and docs and tests that state the current
pin. Recorded 1.4.0 observations, fixtures, CHANGELOG history and the CSO
images' oven/bun:1.3.10 base stay.
* fix: --case runs a single-case file whole
A session option naming the case (testName: 'office-hours-section-loading')
made caseSelection pick test-name mode, whose filter matched no Bun test, so
every trial would run hollow. A file that registers only this case now runs
whole before any name match is tried.
* fix(office-hours): reviewer returns a hash-bound receipt instead of echoing its verdict
The spec reviewer now Writes its verdict once, seals it with
gstack-office-hours-review seal (validates the saved JSON and prints
OFFICE_HOURS_VERDICT round=N sha256=<hash> path=<verdict>), and returns
only that line. check --receipt recomputes the hash and fails closed on a
missing, malformed, stale-round, wrong-path or mismatched receipt. The
saved file stays the sole findings inventory; validators bind the receipt
to the saved bytes and the judge reads those bytes.
* fix(office-hours): only blocking findings send a design back for another review round
Review findings carry severity blocking|minor (verdict schema v2). PASS when
no blocking findings remain; minor findings are recorded in Reviewer
Concerns and the Spec Review table, never a reason for another round.
CONVERGENCE needs a persisting blocking prior. A persisting or unverified
blocking prior cannot be relabeled minor. The 3-round cap is unchanged.
* fix(office-hours): rounds 2 and 3 are delta re-reviews of the captured design diff
prepare saves each reviewed design as round-N.design.md and gives later
reviewers the exact line diff since the previous round. A new later-round
blocking finding must cite (changed_text) an excerpt of a changed line;
seal, check and the validators verify it against the snapshots and fail
closed when a citation or snapshot is missing. Other new observations are
minor. Severity, anti-downgrade and CONVERGENCE rules are unchanged.
* test(office-hours): preservation judge accepts COMPLETED with recorded minor findings
* fix(office-hours): parent fixes only blocking findings; callers may set --max-rounds
Between rounds the parent edits for blocking findings only; minor findings
stay recorded. gstack-office-hours-review takes --max-rounds 1|2|3 (default
3) on every command; reaching it stops at MAX_ITERATIONS. The marathon
fixture sets 2 rounds so round 1 (full) and round 2 (delta) run live.
* docs(changelog): office-hours review loop changes in v1.91.28.0
* fix(cso): helper JSON keeps run URLs only in the validated workflow slot
The shared redactor now treats GitHub Actions run ids as CI metadata
(c245d1bc2), so main's #3041 invariant that the same run digits stay
redacted outside qualification.workflow held only through the phone
false positive. Make it explicit in the CSO sanitizer.
* test(plan-ceo): split-overflow progress detector accepts the skill's candidate wording
Collection never ended when the model wrote 'E1 Slack:' / 'E1-SLACK:' or
labels like 'Include in this quarter' / 'Defer with flip trigger', so the
session ran the whole review (930-1464 s) and failed on its extra findings
or the case deadline (6 of 21 runs). The candidate check now accepts a
platform-bound E-id and the action is the bucket's leading verb, as the
skill names them (Include/Defer/Cut/Hold). The semantic judge, call floor
and include/defer/cut validation are unchanged. Replays the captured CI
37374526437, t2, t9 and t15 menus with negative controls.
* test(plan-ceo): split-overflow candidate identity is the one E1-E5 token
The skill fixes no header or question wording, and shape-matching kept
missing new forms (header 'D1.1 Slack' with 'SCOPE-E1:', 'D1.1 E1 Slack',
'E1 Slack DM bot ...:', a misspelled 'Mattermst'). Identity is now exactly
one E1-E5 token across the header and the question's first line, with no
other candidate's platform named. Options rules, judge, call floor and
include/defer/cut validation are unchanged. Replays all 31 measured runs;
the quoted-row controls now quote a second ledger id.
* v1.91.30.0: move past #3044's claimed 1.91.29.0
* test: register the split-overflow wording replay as a free fixture
* fix(skills): resolve judge-cited ambiguities in /qa and /plan-eng-review
- tasks JSONL keeps the raw branch as GIT_BRANCH instead of overwriting
gstack-slug's sanitized BRANCH used for artifact names
- the Codex outside-voice block runs for ready or unverified, matching
its routing rule
- plan-eng startup names the preamble's own setup prompts; the
cross-project question waits for Step 0
- /qa 8c: a fix that applies but leaves the finding reproducing gets one
revised fix, else it is reverted; browser verification names its before
screenshot; the exploratory loop drops its duplicate 3-step summary
* fix(plan-eng): keep the startup steps in their pinned order
* fix(plan-eng): explicit spawned sessions follow the spawned rule at the scope gate; one announcement rule
A missing tool still sends the prose menu whatever the session type, but a
session the caller or host explicitly marks spawned selects the menu's
Recommendation without prose or BLOCKED. The pre-load announcement
sentence now states Step 0's single check.
* eval policy v3: judge panels gate on the per-dimension median of 3 samples
A judge dimension now passes when at least 2 of its 3 samples meet the
unchanged threshold (judgePanelMedian); any other sample count or a
non-numeric sample fails closed. Thresholds, sample count, judge model and
prompts are unchanged; panel logs show every sample and the mean, and name
the median as the gate. Applies to the skill-llm-eval panels, the
workflow-judge cache, the auq-matrix substance panel and the calibration
replicas. EVAL_POLICY 2 -> 3 and HARNESS_VERSION 1 -> 2 (verdicts change),
so every case starts a new pass-rate series. Replays the captured panels
from jobs 112032744416 and 112049856586 as replayed evidence.
70 KiB
Testing internals: env keys, hermetic E2E
Moved verbatim from CLAUDE.md (token-load reduction). Read this before
writing or debugging E2E tests, passing env: to a runner, or touching
test/helpers/hermetic-env.ts.
Env keys in Conductor workspaces. The GSTACK_* env-shim (v1.39.2.0+,
lib/conductor-env-shim.ts) promotes GSTACK_ANTHROPIC_API_KEY /
GSTACK_OPENAI_API_KEY to their canonical names inside gstack's TS binaries.
Tests run through gstack entrypoints inherit this promotion automatically.
Don't echo the key value to stdout, logs, or shell history. The historical
"never pass env: to runAgentSdkTest" rule is retired: the failure was
partial-env replacement (the SDK's Options.env REPLACES the child's entire
environment, so an object without the key broke auth). The runner now always
passes a COMPLETE hermetic env with per-test env: merged last, so per-test
overrides are safe; ambient process.env.ANTHROPIC_API_KEY mutation also
still works (the env builder reads process.env at call time).
Hermetic local E2E (default). Every E2E runner (claude -p, PTY, Agent
SDK, codex, gemini) spawns children through test/helpers/hermetic-env.ts:
allowlist-scrubbed env (operator CONDUCTOR_*, CLAUDE_*, GSTACK_*,
MCP_*, GBRAIN_*, and credentials like GH_TOKEN never reach children),
a fresh seeded CLAUDE_CONFIG_DIR (no operator ~/.claude CLAUDE.md /
MCP servers / skills), a temp GSTACK_HOME, and --strict-mcp-config.
Local eval signal matches CI. Debug against real operator state with
EVALS_HERMETIC=0 (restores the legacy env AND drops the strict-MCP flag).
Per-test env: overrides merge last, so deliberate contamination
(CONDUCTOR_WORKSPACE_PATH, per-test GSTACK_HOME) keeps working. The
hermetic config dir seeds NO skills by default; a PTY test that types a
/skill slash command must pass seedSkills: true to the PTY runner, which
points the child's CLAUDE_CONFIG_DIR at hermeticSkillsConfigDir() — a
seeded registry that symlinks the LIVE working tree's SKILL.md files (by
design: the skills ARE the subject under test; a snapshot would measure stale
copies). Wiring is pinned by test/hermetic-wiring.test.ts (static tripwire),
two gate-tier canaries in test/skill-e2e-hermetic-canary.test.ts, and the
seeding tripwires in test/hermetic-skills-seeding.test.ts /
test/pty-skill-seeding-wiring.test.ts.
Seeded planning sessions also receive an isolated runtime home through
test/helpers/hermetic-skill-runtime.ts, so absolute lazy-section paths resolve
to the working tree under test. Explicit per-test home overrides remain intact.
Autoplan resolves each review skill from its own installed host registry.
Interactive planning evidence. Native plan-review count drivers use
observeScreen: true and await currentScreen() before choosing an input. The
existing xterm dependency interprets cursor moves and erases; old menus in the
raw stream cannot establish a current prompt. Snapshots preserve
terminal.raw.log, terminal.visible.log, and terminal.screen.log separately.
Setting EVALS_RUN_ID or GSTACK_EVAL_DIR retains these snapshots; an output
directory without a run ID gets a stable, unique local ID for that writer.
Completed native transcript calls establish question counts and phase coverage.
Report-aware count tests also require a fresh, complete report and native
completion evidence before accepting a completion heading.
The periodic first-question matrix uses test/helpers/auq-native-capture.ts
to match the first public PreToolUse AskUserQuestion payload to its current
native display. It grades that question's exact public fields without answering
it or reading model transcripts. question_captured records
workflowCompleted: false. With GSTACK_EVAL_DIR, EVALS_RUN_ID, or an explicit
run ID, native-auq/<run-id>/<test>-<suffix>/capture.json under the eval directory
retains the public payload, bounded current viewport, and capture outcome.
CEO mode selection uses the actual SDK AskUserQuestion permission callback
in auq-sdk-capture.ts, with the existing 12-turn and 240-second limits. It
captures the public question and stops without submitting an answer; its
question_captured outcome also records workflowCompleted: false. The retained
capture survives fixture cleanup. Provider refusals and malformed questions
remain failures. Section-loading captures retain their noninteractive contract.
Shared-code revalidation fixtures pair public tool calls with their successful results to verify that the current trusted start record was inspected before completion. A discovered path in tool output counts; a path mentioned only in instructions or narration does not. Saved public captures cover absolute and relative paths and discovery followed by a read. The revalidation prompt supplies the path to the trusted start-record directory and declares the existing turn limit. It asks the agent to batch independent reads and retrieve the complete final record; every source, approval, persistence, and completion check still applies. The path-boundary fixtures use this same execution contract for symlinks, submodules, ignored files, index flags, and legacy or filtered evidence. Their skip actor accepts an explicit no-change choice; a preservation word inside an option that also approves changes cannot authorize edits. Captured native questions exercise the actual answer callback, and native turn-limit failures still fail even after a question was answered.
The engineering and DX finding fixtures check coverage of their seeded issues rather than cap the total number of review questions. Each decision needs a distinct, completed native question with an offered answer; accepting, rejecting, or deferring a recommendation all count as reviewing it. Engineering's mandatory legacy regression tests also need affirmative plan or public-narration evidence. Additional useful questions are allowed within the existing time limits. Generic question counts remain diagnostic, and a fresh final review report is required.
E2E tests stream progress in real-time (tool-by-tool via --output-format stream-json --verbose). Results are persisted to ~/.gstack/projects/<slug>/evals/ (legacy
fallback ~/.gstack-dev/evals/) with auto-comparison
against the previous finalized run (in-flight _partial files are never used as
a baseline, so a run can't compare against itself).
The periodic overlay fixtures use a versioned behavior gate with efficacy
reported separately. See Overlay benchmark contract v4
for exact correctness requirements, retired fanout cases, immutable evidence,
and the limits of a passing result. Under v4 a completed wrong answer is a
measured correctness result in either arm (taskCorrect: false with
answerError); only an overlay-ON wrong answer fails the case, so an
overlay-OFF wrong answer stays in the comparison. v3 verdicts keep their
original meaning.
Coverage ownership
The test portfolio audit separates deterministic harness checks, prompt judges, live first-question captures, completed workflows and platform integrations. Shared setup is reusable; evidence with different scenario or independent-trial requirements is not. It records the preserved coverage and measured component savings from the test-speed refactor.
Runners: how the suites execute (2026-08 overhaul)
Aside-only E2E tests self-skip without a live Aside; browser-driving tests
run on either engine. Every skill that opens a web page drives the Aside AI
browser first (scripts/resolvers/aside.ts) and falls back to gstack's own
browse engine when Aside is absent (and Chromium bootstrapped). The cases that
need Aside itself (test/skill-e2e-aside.test.ts, design-review-fix in
test/skill-e2e-design.test.ts) call asideAvailable() from
test/helpers/aside-available.ts (the same probe the skills run in BROWSER
SETUP) and skip when the aside CLI or the Aside app is absent. CI runners have
no Aside, so those run only on macOS dev machines and sit in the periodic tier;
set GSTACK_SKIP_ASIDE=1 to force the skip locally (which also exercises the
fallback hand-off). The qa E2E files (test/skill-e2e-qa-workflow.test.ts,
test/skill-e2e-qa-bugs.test.ts) gate on asideAvailable() || fs.existsSync(browseBin): the skill's own BROWSER SETUP picks the engine, so
on a Mac they drive Aside and on Linux CI they drive the built browse binary,
skipping only when neither exists. The $B-driven E2E cases and browse/test/
run on every platform as before, so Linux CI proves the fallback engine live.
Bootstrap dependency retention is opt-in qualification, not the behavior test.
qa-bootstrap still runs its original unpinned Vitest installation and assertions
on macOS and Linux, including documented unsharded commands. Only a Linux paid
shard runner issues the owned retention scope: it binds each actual fixture and
native lifetime, retains locks, package manifests and the installed file/link
inventory, and acknowledges capture before deleting the fixture. The outer
runner also captures evidence when a callback is killed. Incomplete capture
fails qualification and preserves the source fixture as well as partial evidence.
Other platforms explicitly report retention as unavailable and still execute the
native behavior test. A run without the runner-issued scope earns no retained
dependency qualification credit; candidate acceptance requiring that evidence
must use the Linux sharded path and verify every attempt's complete capture,
acknowledgment and cleanup fallback. A passing unsharded or macOS behavior test
does not substitute for that evidence.
The renderer picks the same way, so the render gates are engine-agnostic.
/make-pdf, /diagram, and design previews print and screenshot their local
HTML through lib/aside-render.ts / bin/gstack-render.ts, which render in
Aside when probeAside() says READY and through the browse engine otherwise.
make-pdf's *-gate.test.ts and test/skill-e2e-diagram.test.ts (periodic,
paid) gate on browserAvailable() (test/helpers/browser-available.ts:
asideAvailable() || resolveBrowseBin() !== null) — on a Mac they print
through Aside, on Linux CI through the browse binary bun run build:gates
compiles, and they skip only when neither exists. Only
test/aside-render.test.ts's two live Aside cases (a full round-trip and a
late-readiness --wait-expr poll) are Aside-only: its option
mapping and generated-script pins run everywhere, and its fake-executable cases
drive both engines hermetically (fake aside / browse scripts on PATH pin
probe classification, the stdout contract, loopback-server policy, the timeout
kill, engine choice and the mid-run fallback). test/gstack-render-cli.test.ts
does the same for bin/gstack-render.ts with GSTACK_SKIP_ASIDE=1 and
GSTACK_BROWSE_BIN pointed at a fake daemon that logs every argv line. A green
gate on Linux proves the fallback engine, not Aside; the Mac run is the Aside
evidence.
The browse-binary leg presumes Chromium bootstrapped: resolveBrowseBin()
only checks that the binary (or the find-browse shim) exists, never that
Chromium can launch, so on an install where the best-effort Chromium step
was skipped (GSTACK_SKIP_PLAYWRIGHT=1) or failed, these gates run and fail
at browser launch instead of skipping. Fix the bootstrap (or move the binary
aside) before running them locally; CI always installs Chromium first.
test/dom-dump-hygiene.test.ts is the one free-suite case on the same leg: it
runs lib/dom-dump.js (the rendered-DOM dump /design-review hands the design
detector) inside a real Chromium page through the built browse binary, so it
self-skips when the binary is absent and, because a cold Chromium launch is
load-sensitive on a busy dev box, runs only in CI or on explicit opt-in
(GSTACK_DOM_DUMP_HYGIENE=1).
Free suite (bun run test:free). scripts/test-free-shards.ts runs N
concurrent shard processes (serial within each) with strict-output
classification per shard. Local defaults use the available CPU affinity
(which Bun also bounds by a container's cgroup CPU quota), floored at one and
capped at 16 on Linux and six on macOS and Windows; GSTACK_FREE_JOBS remains
an explicit override.
This does not change the separate CI machine count. Full-suite shards are packed by RECORDED PER-FILE
DURATIONS (LPT, packShardsByDuration) when the committed seed
scripts/free-test-durations.json exists — refresh it with
bun run test:ubicloud --record-durations, which times each file in its own
child on a VM with the CI lane's environment and copies the seed back; see
Free suite duration seed. Missing seed → silent hash-shard fallback; corrupt seed → one warning +
fallback; unknown files get 99th-percentile pessimism, and both full-suite and
--ci-plan runs name them on stderr so a new slow file cannot silently become
the long pole. Packed
shards get duration-aware walls (max(base, predicted × 3, files × 5s)). The
legacy --shards N --shard i path keeps stable hash indices. Required CI uses
one duration-packed --ci-plan, 20 ordinary --ci-run shards plus a separate
exclusive-fixture shard, and a --ci-verify aggregate. CI shards still run on
independent machines without ordering unrelated jobs. The public
TREE_MUTATING map now classifies exclusive host-state fixtures; its sole
entry is test/bootstrap-retention.test.ts, whose same-UID nondumpable actors
affect host-wide procfs permission checks. Locally, this file runs only after
all parallel shards settle, and cancellation prevents that final phase from
starting. No selected files, retries, budgets, or receipt requirements are
removed. Former generator mutators still render into private output directories;
this serial phase protects process visibility, not in-place doc generation.
Full child output is retained in private files under .context/free-test-logs/,
outside each shard's temporary cleanup directory. The runner prints the path at
launch and completion. Losing the log fails the run even when the child exits
successfully. A redirected log directory is rejected before launching a child.
On failure, read that log first: the recovery message distinguishes incomplete
capture, unconfirmed cleanup, deadline expiry and a test/module failure. Fix the
demonstrated cause before rerunning. A focused bun test command is offered only
when every failure is attributable to existing selected files; it proves that
repair, not completion of the original selection. Preserve failed attempts when
sharing results, and inspect logs for private data before sharing them.
Before publication, classify new deterministic regressions for quick feedback.
Refresh the timing seed with the existing recorder on fixed inputs; do not edit
source while tests run. Critical boundary controls belong in QUICK_CORE when
their feedback cost is justified. Other measured files qualify at two seconds
or less; slow and unmeasured files remain outside quick, not outside full tests.
Report cold setup separately from warm execution, while retaining failed-attempt,
retry and cleanup time in the total cost.
PTY fixture timing. Plan-count sessions wake on terminal output or exit,
with at least 250ms between expensive observations and a 2s fallback for
transcript or hook changes that produce no terminal output. New output batches
settle for 250ms before observation so split terminal redraws cannot route input
from their first chunk. Input debounces,
permission guards, and the real CLI's 8s startup grace are unchanged. Synthetic
CLIs can pass startupReadyMarker to runPlanSkillCounting and emit that exact
marker after installing their input handlers; a missing marker fails before
any command is sent. The marker wait stays inside the existing startup and
total-run deadlines. test/pty-output-wake.test.ts covers output, silent waits,
exit, close, missing readiness, split redraws, and continuous redraws. Close and output waits
cancel their losing deadlines so completed workers can exit immediately.
The UI-positive design gate commits the supplied plan as review-input.md
before the first model turn. It requires an acknowledged Design focus/rating
question followed by an answered native finding about concrete UI behavior;
setup and outside-review choices cannot satisfy that sequence. Its declared
board actor submits feedback before acknowledging it. The final proof uses
the full native question, not the truncated diagnostic snippet, and proposal
text mentioning "no UI scope" is not treated as an exit verdict. Unknown-command
failures must name the invoked slash command; a child tool rejecting --help
is not a skill registration failure. Periodic seeded-finding classifiers
separately verify fixture-owned findings.
Paid suite (sharded runner, local AND CI). scripts/test-paid-shards.ts
is the single selection engine: 1 file per shard, EVALS_JOBS shard
processes × EVALS_CONCURRENCY within-shard, per-shard GSTACK_EVAL_DIR,
full-stream spooling to per-shard log files (path printed at START and on
failure), never-started/timed-out taxonomy, and parent-computed diff
selection propagated to children via EVALS_SELECTION_JSON (fail-open: a
child that can't parse it recomputes locally with one warning). Paid evals
never retry; each case's kind fixes its trials before the run (see "Eval verdict
policy" below). Files in CASE_SHARDED_FILES run one registered case per
process (<file>#<case id>, an exact --test-name-pattern, exactly one executed
case), so a long file of short cases spreads across runners and each case gets
its own SDK semaphore. test/carve-section-loading.test.ts is one of them: it registers
one periodic case per carved skill (carve-section-loading-<skill>, each with
its own touchfiles and history). Run one case with
bun run scripts/test-paid-shards.ts --tier periodic --case carve-section-loading-<skill>.
Trial telemetry rides the store: every recorded test carries its 1-based
attempt plus, on an isolated trial shard, its case_id, kind, trial,
panel and policy_version, and each lane's report uploads one
trial-outcomes JSONL line per trial. bun run eval:pass-rates
turns that history into per-case pass rates
(see "Pass-rate history" below; free-suite flakes are a separate record, see
Flake ledger). Census integrity is
enforced from the free suite: every E2E_TOUCHFILES / LLM_JUDGE_TOUCHFILES
key must name a living paid test (test/touchfiles.test.ts's reverse
invariant), and git show <sha>:path fixtures are banned — vendor the bytes
instead (test/git-ref-fixture-tripwire.test.ts).
Functional QA and documentation acceptance require an explicit EVALS_RUN_ID;
GSTACK_EVAL_DIR alone does not satisfy their evidence-ownership guard. For each
local invocation, supply a fresh ID to the documented detached runner:
EVALS_RUN_ID="local-$(bun -e 'console.log(crypto.randomUUID())')" bun run eval:bg:pr
Neither package scripts nor gstack-detach invent this identity. CI's PR/manual
slices, periodic slices and weekly gate census supply an ID bound to the workflow
run, attempt, job and slice. Existing slice artifacts retain per-shard snapshots;
separate always-run native-captures-<EVALS_RUN_ID> artifacts retain project/legacy
e2e-runs and evals/qa-callers evidence for 90 days. These diagnostic artifacts
are not collector results and do not establish that an unfinished test passed.
Fast PR profile and evidence reuse. test:pr selects the changed cases in
scripts/test-pr-profile.ts plus every changed quality judge. --profile full
retains the broad census; no case IDs or tier assignments are removed. The plan
records both E2E and judge selections and lists deferred coverage. Executors
receive those selections separately and the report checks actual executed case
counts. Unknown source dependencies restore the broad gate (see
PR paid lane fallback); unmapped prompts
without registered coverage require full validation. Known broad-only prompt
changes are explicitly deferred, not counted as PR passes. Weekly/manual runs of
evals-periodic.yml execute both complete censuses fresh; manual evals.yml
runs the broad gate. This deliberately moves some defect detection later.
scripts/eval-input-cache.ts accepts only complete, clean, first-attempt passes
with matching before/after inputs. The audited workflow-judge adapter hashes the
actual expanded prompt, source/fixture/rubric/runner closure, installed SDK,
model parameters and runtime. Missing/unknown inputs force execution. Receipts
are scoped to the same repository and PR, expire after 24 hours, and contain
public scores and provenance rather than prompts or secrets. Of the 17 cases
using runWorkflowJudge, 16 are eligible; the cookie workflow's custom input
does not match the cache adapter and stays fresh, as do the other 11 quality cases.
CI supplies the scoped cache/runtime configuration; local runs are fresh by
default. Cached scores must
pass current assertions; reused records retain their original source and time
and cannot renew the receipt. scripts/e2e-shard-reuse.ts extends the same receipts
to PR-profile E2E shards (paid evals never retry, so a pass is structurally a
first attempt): the identity hashes the test's import closure, every tracked file
matched by the touchfiles of every case the file registers plus the global
touchfiles, the runner/workflow/setup actions, the child's environment pins, the
CI image and Claude CLI version, and the shard's case ids, pattern, wall and
concurrency. A computed registration, an unmatched touchfile pattern, a retrying
file, a preload option or a custom endpoint makes the shard ineligible. A reused
shard reports reused with its source run and writes execution: "reused"
collector records; the report rejects reused outcomes outside the fast PR profile.
EVALS_FRESH=1, periodic, marathon and release validation bypass both lookup and publishing.
Free test timing and isolation. test:quick is an explicitly partial measured
subset for edit feedback. test remains complete local acceptance with its
bounded worker pool. The CI planner inventories every free file and packs them
using scripts/free-test-durations.json; each machine runs its assigned shard
serially. Plans and receipts bind the source revision, complete inventory and
strict outcomes; missing, duplicate or mismatched receipts fail the required
aggregate. The existing maximum five-file flaky retry allowance applies across
the entire lane, not separately to every machine. Refresh the full timing list
with bun run test:ubicloud --record-durations. Profiling records failures faithfully
and is separate from final release acceptance.
CI planner/executor/report. --emit-plan <path> --slice-budget S --jobs J
(CI) or --slices K (local) computes selection + the slice plan ONCE (killing
per-slice selector divergence);
--plan <path> --slice i executors consume the manifest and write
slice-result artifacts; --report <dir> reconciles them FAIL-CLOSED (a slice
whose artifact never landed, or a planned shard nobody reported, is a
failure). Budget mode (packBySliceBudget) places shards longest-recorded-first
into the fullest slice whose estimated wall on J FIFO workers stays within S
seconds, else a new slice; a shard with no recorded wall weighs the whole budget
(its own runner), a shard longer than the budget runs alone, and overlays keep
one final one-at-a-time slice. The manifest's plan records each slice estimate
and its own CI job ceiling in sliceCiTimeoutMinutes: max(2 × slice budget, the
slice's longest shard supervised wall, the serialized overlay envelope) + 20
minutes setup (CI_SETUP_ALLOWANCE_MINUTES); ciTimeoutMinutes is the largest of
them. CI derives the matrix ([range(1; .sliceCount + 1)]) from the manifest and
indexes each slice's ceiling as its job timeout
(fromJSON(needs.plan-slices.outputs.slice_timeouts)[matrix.slice]). Executors
refuse an EVALS_JOBS other than the planned J and run their shards longest first.
Each executor's first step records its job start (GSTACK_SLICE_JOB_STARTED_AT).
The slice deadline is that start + the slice ceiling − a 5-minute upload reserve:
after it the executor starts no more shards. A shard that never started reports
not_run (INFRA, never-started); a shard still running is killed and reports
hung (TIMEOUT, timed-out). The executor checkpoints slice-N.json after every
shard start and finish, so a job killed before its final write reports from the
checkpoint (slice N/M ended before its final result …: TIMEOUT), never as a
missing slice. --slices K keeps the supervised round-robin
baseline re-packed by recorded times for local runs. Durations are recorded per
tier (a file's gate and periodic cases differ); refresh one tier from a
downloaded report directory with --report <dir> --write-durations. Under EVALS_ALL the hollow-shard guard marks exit-0 shards with
ZERO executed tests passed-empty (a failure) — census-health, not just
test runs. evals.yml runs the sliced gate lane per PR — the ONLY paid lane
since the legacy 17-row matrix (22.6 min/$21 per PR serialized ahead of the
slices) was deleted after demonstrated parity; its
KNOWN_MATRIX_GAPS/KNOWN_TIER_UNSET ratchets retired with it and
test/evals-workflow-wiring.test.ts pins the surviving wiring (slice-count
agreement, tier consistency, the shared register-skills composite with its
fail-fast verification loop). Tier marathon (complete start-to-finish flows)
is selected positively: a file enters the marathon plan only when it declares
describeE2ETier('marathon') or registers a marathon-tier case, and the gate and
periodic planners exclude marathon-only files; evals-marathon.yml runs them
weekly and on dispatch, fresh, one file per runner, with its own fail-closed
report and tracking issue, and nothing requires it. evals-periodic.yml runs ALL
periodic-tier files weekly (the coverage contract) minus the reasoned
exclusions in test/helpers/periodic-exclude-data.ts (reason + tracking
required per entry; removal re-activates the file), plus a weekly
EVALS_ALL gate census. A red census writes one report on every ref, to the
step summary and the census-report-a<attempt> artifact: the run's branch, sha
and event, both lanes' summaries, every pass-rate alarm line, the red ledger
and session headroom (eval:pass-rates --reds / --headroom --runs 10 --branch <ref>), the all-green estimate and a link to the red-census guide, with
report-derived text passed through scripts/lib/published-text.ts. Only a run
on refs/heads/main comments on (or opens) the one tracking issue and closes it
on the next green run; the marathon lane does the same with its own issue and
marathon-report-a<attempt> artifact. test/evals-tracking-issue.test.ts
pins the main-only guard and runs both report scripts. The host-run Codex job
also carries the live Codex multi-block case (codex-multiblock-live: ./setup --host codex from the job's checkout into a fresh CODEX_HOME, the router in
codex debug prompt-input, and a later /learn block rooted by its own
prelude), and the periodic lane runs the design default-model smoke
(design-model-smoke), a visible SKIPPED with no coverage credit when
OPENAI_API_KEY is absent. The CI
image pins the claude CLI to an exact version (.github/docker/Dockerfile.ci,
enforced by test/ci-image-cli-pin.test.ts — bumps ride PRs that run the PTY
gate), and every eval-store run records claude --version, resolved once in
the runner parent and handed to shard children as GSTACK_CLAUDE_CLI_VERSION
(never spawned on a test thread), so a TUI-drift flake hunt is a grep, not
archaeology.
Eval verdict policy (EVAL_POLICY version 2 in
test/helpers/periodic-exclude-data.ts; version 1 pre-registered 2026-09-29,
version 2 approved 2026-10-04 with the decision memo in
test-audit-2026-10.md). Paid evals
never retry. Each live case has exactly one kind in E2E_KINDS
(test/helpers/touchfiles-data.ts; test/eval-kinds.test.ts enforces coverage),
and the kind fixes its trials before the run:
rule(default): one trial; any failed assertion fails the verdict. For cases where nothing stochastic decides the verdict, or where it checks a contract the product must meet every run.behavior: a panel ofn = 3independent trials, launched together as isolated case shards on different slices (key<file>#<id>~t<N>). All three always run: no early stop and no conditional extra trial. PASS when at leastk = 2pass and no trial violated a contract (expectContract()stampsfailure_class: 'contract'). Each behavior case names its tolerated deviation inBEHAVIOR_WHYand must have a literal registration so it can run alone.judge: an LLM judge scoring a fixed input.judgePanel()(test/helpers/llm-judge.ts) draws 3 samples of the same prompt concurrently inside the unchangedJUDGE_MS; numeric dimensions gate on the per-dimension median of exactly 3 samples, so at least 2 of 3 must meet the unchanged threshold (judgePanelMedian(); no dimension compensates for another, and any other sample count fails closed), booleans on a strict majority. Panel logs show every sample and the mean for information; the median is the gate (EVAL_POLICY v3, approved 2026-10-06). A sample that errors (refusal, truncation, non-JSON, a malformed field) fails the panel and is never resampled; a refusal counts as an unscored panel only when every sample refused.callJudge's 429 backoff happens before any model output and is transport, not a verdict retry. The workflow-judge cache stores whole panels only. Therulecase auq-matrix uses the same panel inside its one execution: each skill's single native capture has its recommendation substance scored by 3 judge samples, gated on their median against the unchanged minimum of 4, and a failed sample fails the panel without resampling (approved 2026-10-04 for censuses after the change).
panelVerdict() (test/helpers/eval-store.ts) is the single verdict
function the report, collector-outcomes.json, the PR comment and pass-rates
all use. A timed-out, crashed or infrastructure-failed trial is a failed trial
recorded with its class; a missing or duplicate trial record makes the verdict
INCOMPLETE, which fails the lane; a 2/3 PASS is shown as PASS 2/3 with the
failed trial's cause. A manual re-run adds trials under a new run attempt and
never replaces the first attempt's verdict. A red census is never rerun on
unchanged inputs: each red is diagnosed as product, test/detector, harness or
infra and resolved by a concrete repair and a fresh census, or listed as a named
red. The one exception: a census whose every red verdict is machine-classified
INFRA or INCOMPLETE (missing slice artifact, runner loss, API error before the
first model turn) may be re-dispatched once as a new run, and both runs are
reported. Changing any EVAL_POLICY constant after seeing census results needs
Garry's re-approval, a version bump and a fresh census;
test/periodic-exclude-policy.test.ts pins the approved values.
Failure causes and details (diagnostic only; verdicts unchanged under v1).
Every failed trial record keeps failure_class (assertion, contract,
timeout, infra: the field verdicts, INFRA re-dispatch and pass rates read)
and its 300-character error exactly as before. failureCauseOf()
(test/helpers/eval-store.ts, called by both record builders, including the
JUnit census for PTY and JUnit-only cases) adds an optional failure_cause
with a one-line failure_cause_evidence. It never changes failure_class.
The highest-precedence cause with evidence wins:
failure_cause |
Recorded when | First step |
|---|---|---|
contract |
expectContract() stamped a contract violation |
fix the product; a violation fails a behavior panel outright |
pre_turn_infra |
the trial failed before the first model turn (failure_class: 'infra') |
check the runner and API status; an all-INFRA census is re-dispatched once |
api_error |
a session ended on an API error (for PTY, the API Error: panel on an idle turn) or the runner's exit reason was error_api |
read the evidence line; a provider outage is a named red |
refusal |
the structured stop reason was a refusal (phrases only as a last resort; quoted refusal text in a tool result does not count) | check whether the fixture prompt invites the refusal |
provider_stall |
a session that streamed partial messages saw no event of any kind, thinking deltas included, for STALL_WINDOW_MS (120 s) while a model request was in flight and no tool, permission prompt, hook or subagent was outstanding; sessions without partial messages are never stalls |
read the session ledger row; still a failed trial under v1 |
session_timeout |
a runner's armed session timeout fired | eval:pass-rates --headroom; cut work, never raise the budget |
observer_timeout |
the test's observer gave up waiting for an outcome (outcome=timeout) |
read the transcript tail: a missed detector or a product that never answered |
assertion |
a check failed after the session ran (session completed; check failed when it ended normally) |
the Expected/Received values or judge dimensions |
unknown |
the trial failed with no evidence for any cause above | inspect the shard directory |
A detector, harness or product fault is a human diagnosis, never a cause.
Readers before this field accept the new records (a frozen v1 reader runs over
them in test/eval-store.test.ts); an unknown cause from a newer gstack is
rejected with unknown failure_cause "<x>" (written by a newer gstack; update this checkout to read it).
failure_detail carries the values error cannot hold: a failed Bun matcher's
{expected, received}, or for a judge one {dimension, mean, threshold, samples, rationale} entry per failing dimension (since EVAL_POLICY v3 mean
holds the panel's gating median), each part capped at 200
characters. sessions[] (at most 32) summarizes the trial's session-ledger
rows. cost_known: false marks a record or trial whose harness captured no
billing: PTY and Codex sessions never bill, and an SDK session bills only with a
terminal total_cost_usd. An absent field means known. Census and marathon
cost lines print the known sum and N of M trial(s) cost unknown, never a bare
total over unknowns.
Session ledger. Every runner (session-runner, the Agent SDK runner
including runSharedInteractive, the PTY session and runRecordedCodexEval)
appends one best-effort line per armed session to
$GSTACK_EVAL_DIR/session-ledger.jsonl (test/helpers/session-ledger.ts):
a stable key <runner>:<label>#n, the exact budget_ms it armed, monotonic
elapsed_ms, the end reason, sanitized evidence, billed, and a compact
liveness summary (longest in-request silence, the last event kind, any tool open
at the end). claude -p sessions stream partial messages for this summary;
transcripts keep their shape and size (full stream diagnostics only with
publicStreamDiagnostics). A failed ledger write never fails a trial; a trial
without ledger rows has no sessions and its headroom is unknown, never
scored. The ledger rides in each slice artifact beside the shard results.
Quarantine (CASE_QUARANTINE, same file). An entry needs: a per-trial rate
below 95% over at least 10 post-policy trials of the case's current input
identity (pre-policy backfill may justify only an initial entry, labeled as
such); a written diagnosis in reason whose failureClass is detector,
harness or model-latency (a product defect is fixed or listed as a named
red, never quarantined); unchanged case touchfiles in the change that adds it;
and an owner, tracking pointer, enteredAt date and measurable exit. A
quarantined case still runs its full panel and reports in every lane but cannot
fail it, except on a hard break (0 of n) or a contract violation, and it never
counts as passing coverage. At most 10% of a blocking tier (gate, periodic) may
be quarantined. The weekly report fails when an entry passes its exit rule (at
least 97% over at least 10 trials) without being removed, when an entry is 8
weekly runs old, or when a tier is over its cap.
Pass-rate history (bun run eval:pass-rates, scripts/eval-flake-rank.ts).
It reads the trial-outcomes artifact of the last N completed weekly
evals-periodic.yml runs, meaning scheduled runs on main plus main
dispatches, plus completed branch census runs in the same window (flags:
--case, --runs N, --branch to inspect one branch, --dir, --backfill,
--json, --gate, and the views --headroom, --reds,
--run <id>; --help works offline, and an unknown flag or case id exits 2
before anything is fetched; every view prints its scope: line). A branch trial counts only toward a series main has also
run (same case-owned bytes, HARNESS_VERSION, model and CLI); a branch never
starts or becomes a case's current series, and weeks for quarantine expiry
count main runs only. It prints
per-case per-trial pass rates with 95% Wilson intervals. A series is one case
under one input identity (Harness version), per model,
Claude CLI version and policy version
(Pass-rate policy versions); a change starts a
new series and older ones stay visible. Labels: INCONCLUSIVE below 10 trials, BROKEN when the latest run is
0/n after a prior interval at or above 95%, FLAKY when failures leave the
interval straddling 95%, FAILING when the whole interval is below it, PASSING
otherwise. --backfill imports legacy slice artifacts as pre-policy trials
(first attempt only; a record that names no registry id is listed as
unattributed, never guessed); they are display-only. --gate (the weekly
report) fails with ACTION REQUIRED, on post-policy trials of the current series
only, when a non-quarantined blocking case meets the entry rule (proposing an
entry), when a rule case does (rule case behaving like behavior: fix or
reclassify), when a blocking case's current identity is significantly below its
previous one (one-sided Fisher exact, α = 0.05, at least 6 trials each side,
Holm-controlled across the cases tested), on the quarantine rules above, and
with a [headroom] alarm when a case's slowest session ran above
HEADROOM_FAIL (85%) of the budget it armed. History that cannot be fetched
fails the gate closed.
The views (scripts/lib/eval-history.ts): --headroom lists each case's
slowest armed session against its budget (warn above HEADROOM_WARN, 75%;
insufficient below 3 samples; timed-out samples marked censored; unknown
without a session ledger, with the case wall beside it as a labelled upper
bound) and the census critical path from the slice job timings. --reds lists
verdict reds per census lane by failure class and cause, pooled and shrunk
per-case red rates, and the all-green probability Π(1 − p_i) over the latest
census's cases with an interval, labelled an approximation because verdicts are
not proven independent. --run <id> is one census's triage: each red with its
values and history, and the transcript evidence it downloads from only the
slices its reds name (a slice artifact over 64 MB is reported, not fetched).
The three thresholds HEADROOM_WARN, HEADROOM_FAIL and STALL_WINDOW_MS
live in test/helpers/eval-budgets.ts, outside EVAL_POLICY; a slow case gets
less work, never a larger budget. The red-census walkthrough is
docs/evals/census-red.md.
The arithmetic. With per-trial pass rate p, the chance a single case goes red (a false red while the product works, the catch rate once it has regressed):
| p | 1 trial | 2-of-3 panel |
|---|---|---|
| 0.99 | 1.0% | 0.03% |
| 0.95 | 5.0% | 0.72% |
| 0.90 | 10.0% | 2.8% |
| 0.70 | 30.0% | 21.6% |
| 0.30 | 70.0% | 78.4% |
The panel removes most false reds at healthy rates, but it catches a 0.95 → 0.70 regression in one run only 21.6% of the time (a single trial 30%, retry-until-green 3%), so drift detection is the history rule's job, not the per-run verdict's. The Fisher alarm is weak at the minimum sample (5.4% power for 0.95 → 0.70 at 6 trials a side), and ten straight passes still leave a 72% Wilson lower bound: after this policy lands, every series starts INCONCLUSIVE.
A lane is all green with probability Π p_rule × Π P(≥2 of 3 | p_behavior) × Π p_judge. For the current registry (PR gate worst case: 107 rule cases and 24 judges; weekly census: 190 rule, 22 behavior and 25 judge verdicts), with rule and judge verdicts at p_rule:
| p_rule | full PR gate | weekly, behavior p = 0.90 | 0.95 | 0.97 |
|---|---|---|---|---|
| 0.99 | 26.8% | 6.2% | 9.8% | 10.9% |
| 0.995 | 51.9% | 18.2% | 29.0% | 32.1% |
| 0.999 | 87.7% | 43.2% | 68.7% | 76.1% |
The rule term dominates: a green lane on a working product needs rule cases to
be near-deterministic (0.999), which is why failing detectors are converted to
outcome checks and product defects are fixed or named, and why each census
reports its all-green probability from the measured rates (eval:pass-rates --reds, carried into the census report).
Timeout policy. Paid tests use the tiers in
test/helpers/eval-budgets.ts (JUDGE/CAPTURE/CAPTURE_LONG/PTY/PTY_LONG);
test/eval-budgets-policy.test.ts pins that every tier fits the shard wall
minus overhead and ratchets raw literals. Budget above the wall is fiction.
No paid test may exceed the ordinary tiers.
FINDING_RETRY_BUDGETS also registers the CEO split-overflow and Eng
multi-finding batching files. Each retains its 25-minute case deadline and runs
once (paid evals never retry) in a 27-minute shard wall including two minutes for
cleanup. No per-case budget grows. Overlay wrappers
have a 1,830-second minimum shard wall and run without Bun retries; see the
overlay contract for their unchanged work budget.
The quality file reserves its whole-file wall (3,170 seconds) for every case run once, plus cleanup. Each still has 120 seconds of model work. Its 17 workflow judges own their deadline and abort signal, with five seconds for terminal recording inside a ten-second Bun grace; the other 11 retain their existing 120-second Bun timeout. Late responses cannot create records or cache passes.
The ship documentation file reserves 4,920 seconds for four 600-second cases and eight 300-second fault cases, run once, plus cleanup; in CI each case runs as its own shard. The standalone documentation child retains its 600-second case. The five review/ship explorer cases reserve 1,695 seconds, run once, including finalization grace. These are whole-file supervision limits, not additional model work per case.
The shared-library path file reserves 1,920 seconds for its three serial
600-second cases, run once, plus 120 seconds for cleanup; in CI each case runs as
its own shard with a 720-second wall (a registered file's case shard supervises
caseMs times its allowed attempts plus the reserve). Its
registered budget keeps the file in its own shard and binds the expected wall
to both the saved plan and the execution receipt; missing or stale budget
records fail reconciliation. Case deadlines and model budgets do not grow.
resolvePaidShardBudget(files, overrideMs?) is the canonical per-job resolver.
Each registered finding file and each overlay wrapper requires its
own shard, even with --files-per-shard above one. Mixed or multi-file overlay
jobs are rejected. An explicit
CLI --timeout, EVALS_SHARD_TIMEOUT_MS, or API timeoutMs still wins for these
policies, including a lower cap; overlay overrides below their minimum are rejected.
Planner entries and execution results record the effective wall,
its source and policy identifier. Custom drivers must resolve each job instead
of passing their ordinary 1800-second default as an explicit cap;
their outer controller/detach wall must also cover the allocated work and cleanup.
The paid census counts are printed by --list for each tier.
eval:bg:<pr|gate|periodic|release> runs scripts/eval-bg.ts. On the local
backend its outer cap is ceil(1.5 × planned serial seconds / EVALS_JOBS) + 20
minutes, at most 4 hours, computed from scripts/paid-test-durations.json for
the lane's current plan (--timeout SECS overrides). On the dispatch backend the
CI job ceilings above bound the run, and the local follower stops after 4 hours.
test/eval-detach-timeout-floor.test.ts pins the cap formula, the 4-hour
ceiling, the backend choice, the sentinel on both backends, the conclusion
mapping and dispatch dedupe.
CI plans with --slice-budget 420 --jobs 2 for the PR gate, the periodic census
and the weekly gate census (the gate census also --skip-judges), and
--slice-budget 1 --jobs 1 (one file per runner) for marathon. The live plans
must fit their workflow's max-parallel so every slice starts at once, and every
per-slice ceiling must stay within 360 minutes; test/evals-workflow-wiring.test.ts
recomputes both from the complete census. Reconciliation rejects missing,
duplicated or misplaced registered work, absent budget records, case shards that
did not execute exactly their case, and reused results outside the PR profile.
Ordinary paid tiers and the default 1800-second shard wall remain unchanged; the
registered and overlay policies above supply exceptions, and unregistered
over-ceiling tests still fail policy checks.
Session timeouts are two-phase: a silent API dies at the startup grace (90s
local / 300s CI floor, distinct exit reason timeout_startup) and the work
budget arms on the first byte — the total wall never grows
(test/session-runner-startup-grace.test.ts pins the floor). A timed-out
session kills its whole detached process group (claude, codex, and gemini
runners alike — test/session-runner-groupkill.test.ts), so a stray
grandchild can't stretch a 600s budget past 1400s. And sync spawns can't
wedge a shard: every spawnSync/execSync/execFileSync/Bun.spawnSync
in the test trees must carry a timeout, enforced by
test/spawnsync-timeout-tripwire.test.ts with a shrink-only exemption
ratchet.
Anchor-sliced setup harnesses. setup is one large bash script, so the
free tests that pin its linker, cleanup, retired-skill prune, browser hint,
rebuild decision, and Chromium-bootstrap behavior never run the whole thing.
They slice the source by anchor (extractFn(name) takes
name() { through the next \n}\n; test/setup-playwright-best-effort.test.ts
slices the # 2. Ensure Playwright's Chromium is available block up to
# 2b.), join the extracted functions with stubbed collaborators, and execute
the REAL bash under a temp HOME with stubbed probes and installers. Two rules
keep the harness honest: renaming a function or anchor comment in setup fails
the test with function not found / anchor not found instead of silently
testing nothing, and test/setup-link-ownership.test.ts and
test/setup-playwright-best-effort.test.ts throw on any command not found on
stderr as harness drift (a helper the test forgot to extract) rather than
letting it degrade into a pass. Files: test/setup-link-ownership.test.ts,
test/setup-cleanup-orphans.test.ts, test/setup-playwright-best-effort.test.ts,
test/setup-prune-stale-generated.test.ts (_prune_stale_generated against a
temp render tree plus host dirs: host cleanup after the generator already
pruned, symlink targets survive, frontmatter-renamed skills, foreign links),
test/setup-browser-hint.test.ts (_browser_hint and the bootstrap summary
across Aside present/absent, bootstrap ok/failed/skipped, GSTACK_SKIP_ASIDE),
and test/setup-needs-build.test.ts (the NEEDS_BUILD block sliced between
two anchors: every binary and source set flips it, Windows .exe suffixes).
test/relink.test.ts shells out to a copy of the real bin/gstack-relink
against a temp GSTACK_INSTALL_DIR / GSTACK_SKILLS_DIR, and
test/hook-scripts.test.ts runs the real careful/bin/check-careful.sh and
freeze/bin/check-freeze.sh with JSON payloads on stdin (including the
GSTACK_HOME state-root parity against bin/gstack-paths).
Pass-rate policy versions
Readers score only trials recorded under their own EVAL_POLICY.version.
Trials from an older policy are shown display-only. Trials from a newer policy
are ignored with a printed count ("ignored N trial(s) recorded under a policy
newer than vX"). When you see that line, run pass-rates from a checkout at or
after the commit that bumped the version. The supported rollback of a policy
change is reverting its writer commit; the reader-compatibility commit before it
stays.
Harness version
Under EVAL_POLICY v2 a case's series identity is the bytes the case owns (its
paid test file, its fixtures and its skills' prompt files, including the
generated SKILL.md) plus HARNESS_VERSION, keyed with model, Claude CLI version
and policy version (scripts/eval-trial-series.ts). Each trial also records the
full consumed-input fingerprint as provenance. scripts/harness-version.json
pins every shared-harness file (GLOBAL_TOUCHFILES plus the session runners, the
PTY harness, the e2e gate and the LLM judge) by blob hash. When
test/harness-version.test.ts fails, run one of:
bun run scripts/bump-harness-version.ts --non-behavioral "<why no verdict can change>" # no series reset
bun run scripts/bump-harness-version.ts --bump "<what behavior changed>" # new series for every case
On a merge conflict in the manifest, take the higher version, then rerun the script.
Ubicloud VMs (bun run test:ubicloud)
bun run test:ubicloud [test:free args] runs the free suite on an ephemeral
Ubicloud VM instead of the local machine. It is the fast path from small dev
boxes, containers, and cloud sandboxes, and the way to record the duration seed
in the CI lane's environment. It needs UBICLOUD_API_KEY (a project token from
the Ubicloud console) plus bash, curl, python3, ssh, ssh-keygen, and
tar locally.
scripts/ubicloud/ubi-runner.sh creates the VM (UBI_SIZE, default
standard-16; UBI_LOCATION, default eu-central-h1) and streams the
checkout to it: tracked files, untracked files that are not ignored, and
.git, so uncommitted edits are tested. It then runs
scripts/ubicloud/setup-free-suite.sh, which mirrors the free-suite CI job
(same Bun pin, Playwright Chromium with its setuid sandbox helper, Xvfb,
poppler, emoji fonts, zsh for the bash+zsh portability arms (#2669), generated
host outputs, gate binaries, and the CSO helper), and runs xvfb-run -a bun run test:free with GSTACK_EXPECT_BINARIES=1
and GSTACK_FREE_RETRY_FLAKY=1. Shard logs are copied to
.context/ubicloud/<timestamp>/, and the VM is destroyed on every exit path.
The exit status is the suite's.
A stock Ubuntu 24.04 VM differs from a GitHub-hosted runner in three ways that
the scripts correct: the login umask is 002 (group-writable directories fail
the CSO private-state checks), AppArmor blocks the unprivileged user
namespaces Chromium's sandbox needs, and clang and python3-venv are absent
(required by the Dia readiness and Python runner tests).
VMs are named ubirun-<epoch>-<hex>. Every new VM first destroys ubirun-*
VMs older than UBI_GC_HOURS (default 12), so an interrupted client cannot
leak one for long. For other commands, use the runner directly:
scripts/ubicloud/ubi-runner.sh run --setup <script> -- '<command>', or its
up / ssh / sync / pull / down steps (--help lists them).
Cloud sandboxes (Vercel / Conductor cloud workspaces)
Syscall-supervised sandboxes need environment setup before bun run test can
run green: run scripts/sandbox-doctor.sh once per boot. It documents and
treats the full failure taxonomy (missing /dev/fd, 64M /dev/shm, spurious
access(2) EACCES from the seccomp supervisor under load, full-capability
processes defeating chmod-denial tests, no X server, no git identity, and
Conductor's git-shim exit-code laundering). The doctor seeds TMPDIR,
DISPLAY, and the runner knobs into ~/.bashrc, so open a new shell (or
source ~/.bashrc) before running the suite. Then:
setpriv --ambient-caps=-all --bounding-set=-all bun run test
Two runner knobs exist for these environments (both no-ops unless set):
GSTACK_FREE_JOBS overrides the shard count in either direction (2 is the measured sweet spot — one
serial mega-shard and 6-way sharding both saturate the per-process syscall
supervisor), and GSTACK_FREE_RETRY_FLAKY=1 re-runs attributed failures once
serially, downgrading a clean retry to a loud FLAKY-PASS (capped at 5 files so
a broken tree can't masquerade as flaky). The required CI free lane and the
Windows lane set the retry knob too, appending every flaky pass to the JSONL ledger it uploads
(GSTACK_FLAKE_LEDGER) — a flaky pass never reds the lane, but it never
disappears either (Flake ledger).
Test selection and tiers
Moved verbatim from CLAUDE.md (#2096 size limit).
Diff-based test selection: the sharded paid runner (test:pr, eval:bg:pr,
test:gate:sharded) selects tests based on git diff against the base branch.
Each test declares its file dependencies in test/helpers/touchfiles.ts. Changes
to global touchfiles (session-runner, eval-store, touchfiles.ts itself) trigger
all tests. Use EVALS_ALL=1 to force all tests. Run eval:select to preview
which tests would run; it applies the PR profile by default, and
--profile full shows the plain touchfile selection. Commands by task: CONTRIBUTING.md.
Two-tier system: Tests are classified as gate or periodic in E2E_TIERS
(in test/helpers/touchfiles.ts — a facade over touchfiles-data.ts +
test-selection.ts). CI runs the changed fast PR profile and selected judges
per PR via evals.yml's sliced lane
(planner manifest → executors → fail-closed report; engine =
scripts/test-paid-shards.ts, the same runner as local eval:bg:pr); the free
suite runs on every PR via .github/workflows/free-tests.yml (a REQUIRED
check, secretless — fork PRs get real signal); ALL periodic tests run weekly
via evals-periodic.yml (EVALS_ALL, minus the reasoned exclusions in
test/helpers/periodic-exclude-data.ts — reason + tracking required per
entry), plus a weekly EVALS_ALL gate census. Use EVALS_TIER=gate or
EVALS_TIER=periodic to filter locally. When adding new E2E tests, classify them:
- Safety guardrail or deterministic functional test? ->
gate - Quality benchmark, Opus model test, or non-deterministic? ->
periodic - Requires external service (Codex, Gemini)? ->
periodic
Tier declarations are enforced by test/e2e-tier-alignment.test.ts (free, runs
in bun test): a skill-e2e-* file named in a touchfiles dep list whose
EVALS_TIER self-gate disagrees with its declared tier in E2E_TIERS fails the
suite. Files not named in any dep list are reported, not enforced — keep both
in sync.
PR paid lane fallback
PR selection (scripts/test-pr-profile.ts) classifies every changed file. A file
mapped by a paid case's touchfiles selects that case. These files are exempt and
select nothing: docs, any non-paid *.test.ts(x), FREE_ONLY_PR_FILES (the
free and paid duration seeds, scripts/ubicloud/**, every workflow that runs no
paid evals, and free-only tools), and the fixtures in FREE_FIXTURES
(test/helpers/free-fixtures-data.ts, each fixture with its free consumers).
PAID_WORKFLOW_FILES (evals.yml, evals-periodic.yml, evals-marathon.yml)
and any other unknown file restore the full gate (prCoverage.mode = full-fallback). The job summary and PR comment list each file that caused a
fallback with one of three labels:
- needs touchfile entry: a paid test, helper or fixture no case maps; add it
to the owning case in
E2E_TOUCHFILES(test/helpers/touchfiles-data.ts). - add to FREE_ONLY_PR_FILES: a file only free lanes read; add it to the list
in
scripts/test-pr-profile.ts(or a fixture toFREE_FIXTURES). - real unknown dependency: shared code (bin/, lib/, resolvers, hosts, setup) whose dependents are not mapped; the full gate is the correct result.
test/free-fixtures.test.ts holds the inventory: each test/fixtures/** path
belongs to a paid touchfile or FREE_FIXTURES, and each workflow to the paid or
free-only list. Editing a paid test file runs its own gate cases on the PR when
their Bun test is named by the case id; the file's other cases stay deferred with
a reason that says to name the test by its case id.
PR receipts, recovery and debounce
Passes and failures of single-trial shards are written as receipts at the
execution boundary, so a cancelled run's finished work survives. The
recover-receipts job runs scripts/recover-receipts.ts from the base ref with
actions: read: it merges the receipts of at most 5 completed, cancelled
evals.yml runs of the same PR (oldest first, negatives before passes, so a newer
FAIL blocks an older PASS) and hands the store to plan-slices, which keeps
contents: read only. Its recovery.json checkpoint never advances past an
unresolved run. plan-slices turns reuse off when recovery degraded, the
checkpoint is missing or a cancelled run is still unresolved; recovery itself
never fails the run, and the job summary names the reason (rate limit with its
reset time, gh auth status, the 60-second budget). The whole store is saved by
slices-report.
On a synchronize push, when another run for the same PR started in the last 15
minutes, the debounce job waits 90 seconds before planning; a run that outlives
the wait reports "superseded by ". PR labels: evals-no-debounce skips the
wait and evals-fresh turns receipt reuse off. Labels apply from the next push,
because adding a label does not start evals.yml.
Free suite runner, judge reuse and engine skips
Moved verbatim from CLAUDE.md (#2096 size limit).
bun run test routes through scripts/test-free-shards.ts (N concurrent
shard processes, serial within each, packed by recorded per-file durations
when scripts/free-test-durations.json exists — refresh it with
bun run test:ubicloud --record-durations; strict-output classification per
shard: a shard without bun's terminal summary line FAILS — silent truncation
cannot report green). TREE_MUTATING lists the files that still run in their
own trailing serial shard (today only test/bootstrap-retention.test.ts, for
host-wide procfs visibility); gen-skill-docs tests no longer need it (main()
guard, and --out-dir renders every host into mkdtemps — see
docs/TESTING_INTERNALS.md). Never type bare bun test for the suite: it
walks the whole repo, loading paid eval files and missing the strict
classifier.
It covers skill validation, gen-skill-docs quality checks, browse
integration tests, the Aside contract pins, and the render-wrapper pins.
bun run test:pr runs the selected short live behaviors and quality judges.
It reports deferred broad coverage; unknown dependencies restore the full gate,
and an unmapped prompt without registered coverage blocks planning. Full free
acceptance and required PR checks must pass before publishing. CI can reuse the
16 workflow-judge passes for 24 hours when their complete consumed inputs and
runtime match; records preserve original provenance. The cookie workflow's custom
input, the other 11 judge cases, dynamic agent tests, and local runs without
scoped cache configuration stay fresh.
Scheduled/manual full coverage and test:release always run fresh.
See testing policy for commands and measured targets.
Anything that needs Aside
itself (test/skill-e2e-aside.test.ts, the Aside qa/design E2E cases, the
live render in test/aside-render.test.ts) runs only on a Mac with the Aside
app open and self-skips elsewhere (asideAvailable()). make-pdf's render
gates and test/skill-e2e-diagram.test.ts run through whichever engine
resolves (browserAvailable() — Aside, or the browse binary CI builds with
bun run build:gates) and skip only when neither exists; the fallback
engine's own tests run everywhere.
Real-home tripwire
No free test may write the developer's real home. scripts/lib/free-home-guard.ts
snapshots ~/.gstack, ~/.claude, ~/.codex, ~/.agents and ~/.config/gstack
(live session logs excluded) and fails the run when an entry changes.
- A shard that runs alone (
--shard, CI, the exclusive host-state shard, the flaky retry) owns its window, so the failure names its files. - Full-suite shards run concurrently on one HOME and cannot tell whose write a change was. The runner guards that whole phase once and names no shard.
bun run scripts/test-free-shards.ts --attribute-homenames the writer: it runs every file alone with a private HOME (browser cache and git identity still come from the real home) and reports each file that wrote a watched surface. The retained shard directory keeps the written files as evidence.
Fix a writer with usePrivateStateRoot() (test/helpers/private-state-root.ts)
or a child HOME/GSTACK_HOME, and resolve product state paths at write time,
never at import.
Flake ledger
GSTACK_FREE_RETRY_FLAKY=1 reruns a failing free test file once, serially. A
clean retry turns the shard green as a flaky pass and appends an entry to that
shard's flake ledger, uploaded as flake-ledger-<shard> (90-day retention). The
Monday "Weekly test health" workflow (bun run test:health --enforce) reads
main's ledgers through scripts/lib/ci-history.ts and fails when a file flakes
in more than 5% of main free-tests runs over a window of at least 20 runs; a
window thinner than 20 runs in the last 7 days widens to the newest 20 runs, and
fewer than 20 runs is report-only. bun run test:health flake-summary <dir>
prints a run's ledgers as a table. A flaky file is fixed at source; the retry
never masks it. bun run eval:pass-rates lists only the local ledger
(flakeLedgerPath(), overridden by GSTACK_FLAKE_LEDGER).
The aggregate job of free-tests.yml downloads every flake-ledger-* artifact
of the run and appends bun run scripts/test-health-report.ts flake-summary <dir>
to its job summary. The Windows lane uploads flake-ledger-windows-<shard>.
Free-suite flake rules:
- Bun gives a child
stdio: 'pipe'a unix socketpair. A test that must hold a write blocked uses a FIFO the test never reads (test/qa-evidence.test.ts,test/qa-deadline.test.ts), not a paused child stream: socket capacity followsnet.core.{w,r}mem_defaultand the parent's reader. - A test that rewrites CI provenance variables (
CI,GITHUB_*,CI_*) restores them. Free shards run many files in one process, and a later browse daemon withoutCI=truelaunches Chromium sandboxed. - Handshakes with a runner wait for the runner's own observable effect (its progress line, a coalesced read count), not for a fixture side file or a timer.
Free suite duration seed
scripts/free-test-durations.json drives shard packing; files missing from it
pack at the seed's 99th percentile. Acceptable seeds come from
bun run test:ubicloud --record-durations (each file timed in its own child on a
VM with the CI lane's environment) or a CI recording. A 4-core laptop or
small-sandbox recording is not acceptable: browser and display tests skip or slow
down there, so the seed misstates them.
--ci-planwarns in the job summary, with the refresh command, when more than 5 files are unseeded; it never fails a PR.--ci-verifylists shards that ran longer than 1.5× their prediction and at least 30 seconds over.- The weekly test-health run fails its enforced
unseeded-free-filescheck ("stale seed: N unseeded file(s) > 5") when more than 5 free test files are missing from the seed, so contributors without Ubicloud access learn within a week that it needs a refresh. test/free-seed-ratchet.test.ts(run with full history infree-tests.yml'sfree-planjob,GSTACK_FREE_SEED_BASE=$(git merge-base HEAD origin/main)) fails when a seeded file over 60 seconds is missing fromscripts/free-test-seed-allowlist.json, when an allowlist entry is stale (gone, or no longer over 60 seconds), or when an entry is added relative to the merge-base copy without atodo. An annotated exception is{file, reason, todo}and the test prints it. A merge-base without the allowlist file counts as the first allowlist. Fix a failure by splitting the file, then rerunbun run test:ubicloud --record-durations.
Windows free lane
windows-free-tests.yml plans on Linux (--windows-only --ci-plan --shards 6),
runs one strict serial shard per windows-latest job
(bun run test:windows --ci-run), and verifies every shard receipt on Linux
(--windows-only --ci-verify). Packing uses Windows-measured durations from
scripts/free-test-durations-windows.json; refresh them with
gh workflow run windows-free-tests.yml --ref <branch> -f record_durations=true,
then download the free-test-durations-windows artifact and commit it.
Curation lives in scripts/lib/windows-curation.ts. Every Windows plan and
--list prints the curated count and the top exclusion reasons (also in the job
summary). Files a probe proved safe despite a /tmp literal are force-included
(WINDOWS_PROBE_SAFE). test/windows-native-workflows.test.ts holds
WINDOWS_CURATED_FLOOR; when Windows-safe tests are deleted on purpose, lower it
in the same PR with a reason.
Native qualification campaigns (Windows cookie extraction, Windows launch
diagnostics, Dia on macOS) run only on dispatch of native-qualification.yml:
gh workflow run native-qualification.yml --ref <branch> -f mode=<cookie-native|windows-native-diagnostics|dia-native|dia-launch-comparison|dia-gui-readiness>
Real-setup install fixture
Tests that run the real setup use test/helpers/install-fixture.ts. Each test
file builds one read-only seed checkout and clones it per test with
copy-on-write file copies (plain copy fallback), never hard links. Writes go
through put, setVersion and the fixture*Sync helpers; files register
afterEach(cleanupFixtures) and afterAll(cleanupSeed).
test/install-fixture.test.ts is the isolation tripwire: writes in one clone
leave the seed and sibling clones unchanged.
Test-only timing knobs
Defaults are the product's; tests shrink them instead of waiting real time:
GSTACK_RENDER_SLACK_MS (lib/aside-render.ts, default 10000),
GSTACK_MEMORABLE_LOCK_TRIES (bin/gstack-memorable, default 50 tries of 0.1 s),
the recordGraceMs option of test/helpers/office-hours-attempt.ts (default
5000) and the timeoutMs argument of clearCookieTargetStorage(page, origin, timeoutMs) (default 15000).
Running evals as an agent: detach
Moved verbatim from CLAUDE.md (#2096 size limit). The short rule stays in CLAUDE.md; this is the full mechanism.
When you (an agent/harness) launch a long eval/benchmark run, run it through
bin/gstack-detach — NEVER as a plain backgrounded Bash task. A plain background
task lives in the harness's process group, so a SIGTERM ("polite quit") on a turn
boundary, a stopped Monitor, or an interruption kills the run mid-flight (observed:
script "test:gate" was terminated by signal SIGTERM ~40 min into a run of the
since-retired single-process runner). On macOS
the run can also die to idle-sleep. gstack-detach fixes both: a fresh session
(escapes the group SIGTERM) wrapped in caffeinate -i (blocks idle-sleep).
- Use
bun run eval:bg:<pr|gate|periodic|release>(scripts/eval-bg.ts). It prints[eval-bg] backend=<dispatch|local> …andgstack-detach LOG <path>and returns at once. Both backends run undergstack-detachwith the machine-widegstack-evalslock (concurrent worktrees serialize instead of saturating the shared model API) and a run-scoped log under~/.gstack-dev/eval-runs/whose first line names the backend, the run and the tested revision.- Dispatch (default when HEAD is clean and equal to its pushed branch on
garrytan/gstack, checked withgh api, andghcan dispatch):prandgatedispatchevals.yml(evals_all=falsewithbase_ref/base_shaforpr,evals_all=trueforgate),periodicdispatchesevals-periodic.yml,releasedispatches both. Every dispatch passesexpected_sha(the workflow fails fast when the checked-out revision differs) and anoncethat appears in the run name (E2E Evals [eval-bg <nonce>]), which is how the log finds and follows its run. A queued, running or green dispatch with the same revision and inputs is followed instead of dispatched again. Dispatched runs are validation runs: fresh and never writing PR receipts, except that aprdispatch on a branch with an open same-repo PR passespr_receipts=<PR>and reads that PR's receipts read-only. - Local (dirty tree, unpushed commits, a fork, no
gh, or--local/GSTACK_EVAL_BG_MODE=local): the sharded paid runner (scripts/test-paid-shards.ts) on this machine, capped at ceil(1.5 × planned serial seconds /EVALS_JOBS) + 20 minutes, at most 4 hours (--timeout SECSoverrides). One Bun process per test file, an external wall-clock timeout that kills the shard's process GROUP (strayclaude/codexgrandchildren included), a per-shardGSTACK_EVAL_DIR=<evalDir>/shards/<slug>/honored by theEvalCollectorconstructor, and an aggregate that separates failed vs timed-out vs never-started shards.EVALS_JOBSsets the shard process count (default 8, 2 forpr);EVALS_CONCURRENCYis bun's --max-concurrency WITHIN a shard (default 2). ExportANTHROPIC_API_KEYfirst (never pass keys in argv). --dispatchrefuses with the fix ("commit and push, or rerun with --local") instead of falling back.eval:list/eval:compare/eval:summary/eval:pass-ratesread the shard dirs too. For any other long agent job, callgstack-detach [--lock NAME] [--timeout SECS] [--label LBL] -- <cmd>directly.
- Dispatch (default when HEAD is clean and equal to its pushed branch on
- Then poll the printed logfile with a death-aware watcher: break on the
guaranteed
### gstack-detach EXIT=<code> ###sentinel (0 passed, 1 failed, 130 cancelled, 2 other on both backends; silence is never mistaken for success).bun run scripts/eval-bg.ts status <log-or-run-id>reconnects and prints running, passed, failed, cancelled or incomplete (died without a sentinel). The detached run survives even if your watcher gets reaped, so re-checking the log always works. Keep checking until the sentinel appears or the user tells you to stop; a long run is expected, and a promise to check later is not a result. At each check, report which tests passed, which are still running, and any failures so far. - Why the lock: a shared dev box with several Conductor worktrees will rate-limit the model API if two eval suites run at once (15-way concurrency each), which mass-times-out E2E tests. The lock makes the second run WAIT, not collide.
- Humans running
bun run test:prforeground in their own terminal don't need this — Ctrl-C is intended there. Detachment is for agent-launched runs only.