On #210 the review and comment JSON outgrew ARG_MAX, and
`jq -n --argjson ...` failed to exec with exit 126 before printing
anything. A `combine` helper now pipes the three documents to `jq -s`
through the printf builtin, on both the fixture path and the live path.
The new suite cases put about 2 MB on each surface, so a fix that moves
only one document off argv still fails. Both rivals fail 5 of 30: the
pre-fix script, and a partial fix that keeps the issue comments on argv.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
#297's rulings kept: no bash floor, no find_modern_bash probe in
checkers.sh (it runs suites under the dev shell's $BASH), and no
toolchain-policing step in the build job. This branch's CI table kept:
ci_suites, --ci, --would-run, and the --ci steps in ci.yml (build, js, kb),
which replace gh/main's hand-listed kb, --self-test and remaining-checkers
steps. The kb row already runs every suite those steps named.
check-ci-graph passes on the merged ci.yml.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
Copilot on #297: `pr-reviews` only calls `date +%s`, so the REAL_DATE
pass-through never ran. It is deleted rather than tested. The fake now refuses
any other arguments with exit 2, so a new `date` call shows up as a failure.
Rival: with pr-reviews calling `date -u +%s`, "a review arriving on the third
poll is seen" and "--wait keeps polling until its deadline" fail (23/25).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
This follows the user's rulings: CI should not care which bash or which
standard utilities it runs, and jobs take what the ci dev shell provides.
- checkers.sh: remove find_modern_bash, the 5.1 floor (improvise-fd0j) and
exit 3. Suites run under $BASH with PATH unchanged.
- checkers-test.sh: remove sections 2, 5 and 5b (the fake bashes and utilbin).
They existed only to pin the floor.
- scripts/kb: drop the BSD `stat -f` fallback. kb needs the dev shell for tpl,
and that shell's stat is GNU.
- ci.yml: remove every `shell: bash` override, so every step runs in the dev
shell. Remove the web-toolchain version probe and the Playwright
version/browser probe. The shell's PLAYWRIGHT_BROWSERS_PATH pins the
browsers. npm ci stays because @playwright/test is not nix-provided.
- flake.nix: update the comments that promised those CI checks.
Reasoning: context/code/tooling.md section 5. Evidence:
context/test-evidence/tooling.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
On the NixOS runners /bin holds only sh. With no /bin/sleep, the watchdog in
bounded_reviews exits at once and wins the `wait -n`, so every --wait case
reports 124. Both tools are now resolved with `type -P` before any case stubs
PATH.
Red: with both paths pointed at /nonexistent/, the suite fails 25 of 25.
Rival: REAL_SLEEP=/nonexistent/sleep fails 25 of 25 again. The date
pass-through has no discriminating rival, because pr-reviews only calls
`date +%s`. Evidence: context/test-evidence/tooling.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
ci.yml: this branch's check-harness-lock --allow-network step is kept, ahead
of main's nix-build clippy and test steps that #228's merge brought in.
test-evidence/tooling.md: both EOF appends kept, no heading lost (42 = union).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
ci.yml: main's NixOS-runner rewrite, with this branch's --ci build and --ci kb
steps. Main's separate `checkers.sh kb` step is dropped (the kb row runs it),
and so is the `nix shell nixpkgs#trealla` wrapper (the dev shell supplies tpl).
test-evidence/tooling.md: both EOF appends kept, no heading lost (41 = union).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
Section 5b (the bash 5.1 floor, #237) was written on a branch cut before the
NixOS-runner change (af44653f). That change moved section 5 off /usr/bin:/bin,
because on NixOS that path holds only env and sh. The two merged with no
conflict. On NixOS the runner then dies on `dirname` with exit 1, which turned
the KB job red on every PR at 666de199.
5b now uses utilbin too, and utilbin gains head and env for the 5.1 control.
Evidence and rivals: context/test-evidence/tooling.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
Main's PR #287 added a workflow-level defaults shell, and the refusal made every
suite read as unrun. The two cases that pinned it are deleted; one shows a step
under workflow and job defaults is counted.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
Recomposes #245 onto #228's CI table. The kb job's checker step no
longer lists suites by hand: it runs `checkers.sh --ci kb`, and
check-ci-graph.py holds the ci_suites table to `checkers.sh --list`.
So check-harness-lock joins the table's kb row, with the one-line
reason that the old ci.yml comment carried. The conflict in ci.yml is
resolved to #228's step. The --allow-network step is unchanged.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
The gate split on \n alone while YAML also breaks at U+0085, U+2028 and
U+2029, so an if: after one was invisible and the gate exited 0. Every line,
comments and block-scalar bodies included, is now checked for C0 (but tab),
DEL, C1 and U+2028/2029. CR is folded by the one read path, read_workflow, and
file-based controls pin that. Argument: context/code/tooling.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
A multi-line flow value (- name: [a, "b ... run: ... c: x"]) is one string to
go-yaml, PyYAML and Psych, but the gate counted the inner run:. Rather than
refuse one more construct, values are now an allowlist: empty, a block header,
a closed quote, a one-line flow with no quotes or nesting, or a plain scalar
not starting with an indicator. The real ci.yml needs nothing else.
Argument: context/code/tooling.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
The tracker closed a multi-line quote on an escaped \" or a doubled '', so a
later quote reopened it over a job's real if:. Both styles exited 0 on the real
ci.yml. The opening line is refused instead; _opens_quote's escape handling
is now load-bearing and has a case per style. Argument: context/code/tooling.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
A job header with a trailing comment was accepted by the readability check
(comment-stripped line) and never opened (raw line), so its if: went to the
previous job and read_graph never listed it. The audit found the stripper also
cut a quoted "x # y" open, hiding a later job-level if:. Both exited 0 on the
real ci.yml. Refusing the comment makes stripped == raw on every accepted
line, so the stripper is deleted. Argument: context/code/tooling.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
w0j4 read any line deeper than 4 outside steps:/needs: as content, so a
needed job whose own keys sat at indent 5 or 6 (valid YAML) had its if:
and needs: missed and the gate exited 0. A deeper line is now content
only once its job has a 4-space key. tooling.md names the three valid
styles the gate refuses loudly.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
improvise-cota, found in the delta review of #244. The tdfd fix skipped an
absent tracked path whenever `ls-files -d` listed it. `ls-files -d` also
lists an absent --assume-unchanged path. `status` hides that path, and
`commit -a` never records the deletion, so its citations stayed in HEAD
unread and the audit exited 0.
The deletion oracle is now `diff --name-only --diff-filter=D`. It honours
both index flags and still lists a plain unstaged delete.
The new case in bd-audit-ids-test.sh was red at 33d563a9. A rival that
treats every absent path as unread fails only the deliberate
deleted-file control.
Stacked on #244 (tooling/lqs0-audit-exits).
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
Quoted, anchored and multi-line needs: values, and quoted, spaced or
explicit ? keys, all skipped kb on a PR while the gate reported every
suite run. Rather than parse more YAML, every structural line inside
jobs: must take one plain form or the gate names it and fails. Also a
1be2 case at the one-column boundary, which the owner+1 rival passed.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
A skip-worktree or sparse path is absent without being deleted, and
the filter skipped it, so its ids were never read and the audit
passed. The xargs comment no longer claims 124, which is GNU only.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
find_modern_bash accepted any 5.x, so a bash 5.0 was pinned and
pr-reviews-test.sh then failed on an invalid option. The probe now
reads major and minor; a 5.0 case and a 5.1 control guard it.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
A plain scalar folds a deeper-indented next line into run:, so
`|| true` there made the counted call fail-open (improvise-1be2).
A job that needs a job under if: is skipped with it, directly or
transitively (improvise-bbwa). The unrun message now points at the
ci_suites row. tooling.md names the checkout-with-ref hole.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
improvise-lqs0, from the #226 review. xargs reports grep's no-match and
its read error alike, as 123. So a cited file the scan could not read
passed the audit, and an unreadable file holding the only dead id exited
0. Each grep batch now maps exit 1 to success and exit 2 to 255, so xargs
stops and the audit is exit 4. A tracked file deleted from the working
tree is left out on purpose, because its ids are leaving the tree.
Two exit-code mismatches with the stated contract are fixed too. Running
outside a repository is now exit 4, not 1. A bd list whose objects carry
no string id is now exit 3, not "every id dead".
Six new cases reproduced the bug as 10 passed, 4 failed, and now pass
14/14. Four rivals each fail their own case.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
improvise-st8q, improvise-tvte, from the #228 re-review at c1417359. The
maax parser still failed open. A bare checkers.sh inside a heredoc or a
folded run: value counted as every suite, and step and job if: were
ignored. Every round of parsing met another shell or YAML form, so the
parsing is gone.
checkers.sh holds a job-to-suites table (ci_suites), and each CI job
makes one call, `checkers.sh --ci <job>`. check-ci-graph.py counts a step
only when all of these hold: it sits inside steps: at step-key
indentation, its whole run: value is that call, the call names its own
job, and no if:, continue-on-error:, shell: or defaults: covers it. Every
other mention of scripts/checkers.sh is reported. A key with an inline
value owns deeper lines, and an open quote owns lines until it closes, so
heredocs, folds and quoted values are content, never structure.
There are 19 suite self-test cases and 7 rivals, each failing a case only
it can. The reviewer's heredoc, fold and if: attacks on the real ci.yml
now exit 1. checkers-test.sh gains five --ci cases. context/code/tooling.md
corrects two overstated claims ("--would-run anywhere is a usage error",
"fails closed").
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
improvise-7fl8: on the one-tick-per-read clock --wait 5 polls exactly 6
times, but the test asserted at least 5. So a loop capped at 5 polls, or
a deadline of now + WAIT - 1, passed at 25/25. It now asserts
polls = WAIT + 1, and both rivals fail it.
With the deadline check deleted, the suite used to hang at the first
real-clock timeout case. Every pr-reviews call now runs beside a
/bin/sleep watchdog (wait -n -p), and loses to it as exit 124. The same
rival now finishes in 93s with four FAILs.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
The #228 review and Codex's P1 found the suite check failing open. A bare
scripts/checkers.sh after an inline comment, in a step name or as echo's
argument read as "every suite runs". --list alone read as every suite, and
--self-test's discarded names, and --list's, counted as run.
checkers.sh gains --would-run, honoured only as the first argument. It
reports what the rest of the arguments would run, from the real argument
loop and name resolution. check-ci-graph.py only finds invocations: in run:
text, comments stripped by bash's rule, command position, every command on
the line, quotes removed. It then asks --would-run about each. An
unparseable line contributes nothing and fails closed.
16 suite self-test cases, 8 of which the old parser gets wrong. Six
rivals each fail a case only they can. checkers-test.sh gains four
--would-run cases.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
improvise-dak5: ci.yml names its checker suites by hand, so a new
scripts/X-test.sh ran locally and never in CI. check-ci-graph.py now
asks `checkers.sh --list` for the suites, so the naming rule keeps one
home, and reads every checkers.sh invocation in ci.yml. A bare run
covers every suite, --self-test covers `checkers`, and a YAML comment is
never a run.
A suite planted in the real tree fails the check, naming it, and adding
it to the step clears it. Seven new self-test cases. Rivals: counting
comments fails the comment case; answering "everything runs" fails five.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
improvise-52eq: the AGENTS.md recipe searched `bd show` output for "not
found" and printed "all resolve" when there was none, so a bd that was
missing, failing or never run passed the audit. scripts/bd-audit-ids
keeps the file scan and exclusions, then `comm -23`s the cited ids
against `bd list --all --limit 0 --json`. A dead id is exit 1 and is
named. A bd that does not answer is exit 3, never a pass.
scripts/bd-audit-ids-test.sh has eight cases on a throwaway repo with a
fake bd. The old recipe as a rival fails 7 of the 8. Disabling the
comparison fails the dead-id case, and an empty exclusion list fails the
pass case. CI runs it with the other named suites. The "One environment
note" now reports which forms the guard refused and allowed, as observed.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
improvise-7epn: two cases on the one-tick-per-read clock. One resolves on
poll 3, and one times out after polling up to the deadline. Replacing the
post-poll deadline check with `break` now fails exactly those two
(23 passed, 2 failed), where before it left the suite 23/23.
improvise-lnsh: when `cargo metadata --locked` refuses the opfs-contract
lock, build.sh now says not to remove --locked and names
`cargo update -w --manifest-path harness/opfs-contract/Cargo.toml`.
Checked by hand against a lock with pest pinned to 2.9.1. Evidence is in
context/test-evidence/tooling.md.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
improvise-qnpi: `pr-reviews --wait` checked its deadline before the first
re-poll, so a second boundary between two reads of the clock skipped the
only poll (exit 1 with polls=1, reproduced under CPU load). The deadline
is now checked after each poll. A new case drives it with a PATH shim
whose `date` ticks one second per read, so it fails deterministically on
the old loop rather than under load.
improvise-mkbp: harness/opfs-contract/Cargo.lock pinned pest 2.9.1 while
its path deps require 2.9.2, so every build rewrote it. The resolved lock
is committed and build.sh starts with `cargo metadata --locked`; the
wasm-pack `-- --locked` route was tried and does not work (wasm-pack
re-resolves first). opfs.spec.js now names the fix when harness/serve/opfs
is missing, which is what running web/assemble.sh alone leaves.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
The required Browser harness check comes from a job that needs: build, so for
a moment per push every registered check is green and GitHub says
MERGEABLE/BLOCKED. #186 at 128a2a0 refused that and exited 1 on healthy PRs.
BLOCKED with no failed check is now waited on (exit 3 if it never clears).
Adds the gap sequence from PR #189's real rollup, and a head moving during
the UNKNOWN wait. Corrects tooling.md's claim that nothing else produces
BLOCKED, the evidence entry's rival-2 claims, and the fixture wording.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
GitHub reports UNKNOWN for a while after the base moves, and PRs #172 and
#179 both exited 0 there. UNKNOWN is now waited on (exit 3 on timeout), a
failed check or CONFLICTING ends the wait, and the pass condition is
MERGEABLE with CLEAN, UNSTABLE or HAS_HOOKS. BLOCKED, DIRTY, BEHIND, DRAFT
and any unshipped state fail closed. The suite enumerates the whole table.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
The kb skill (both copies) now describes the HEAD comparison, the ledger,
exit code 3 and the recovery. kb-tool.md records the ih3k/yea8 arguments and
corrects the residue paragraph: .gitignore hides the leftovers, and there are
six, not one.
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VdfEGXA7aJ6ftMsCc9PJze
Edward Langley