Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 10 additions & 4 deletions crates/squabble-core/data/first-line-rules.tsv
Original file line number Diff line number Diff line change
Expand Up @@ -18,26 +18,28 @@
# A new rule is added here and to ADR 0004 together, and the class counts in
# the catalogue_shape test are updated with it.
SQ-LOCK-01 FIX invalid lockfile pin: an actions.lock record keyed owner/repo@<40-hex> repeats that SHA as its ref Refresh scripts/check-lock-sync.sh from template main, then set each flagged ref to the highest full-semver tag at the SHA, else highest semver tag, else first tag, else a branch containing the commit The template lock-sync gate exits 0 and the failed workflow no longer reports invalid lockfile pin; scripts/check-lock-sync.sh needs GNU awk and mikefarah yq v4, and a missing tool means the check did not run, never a pass No tag or branch holds the SHA; tags tie; the lock also lacks entries
SQ-LOCK-02 FIX lockfile missing pin for an action called inside another action; or lockfile verification did not produce a result because a locked ref no longer resolves (unable to find version <sha>); or a workflow bumped an action past its lock (stale lock, found statically) Regenerate the lock with gh actions-lock The lock-sync gate passes and the job starts The regenerated lock disagrees with the workflow YAML
SQ-LOCK-02 FIX lockfile missing pin for an action called inside another action; or lockfile verification did not produce a result because a locked ref no longer resolves (unable to find version <sha>); or refs missing from the lockfile (check-lock-sync.sh: FAIL <workflow>: refs missing from the lockfile: <action>@<sha>); or a workflow bumped an action past its lock (stale lock, found statically) Regenerate the lock with gh actions-lock The lock-sync gate passes and the job starts The regenerated lock disagrees with the workflow YAML, or the missing ref is an action called inside a reusable workflow that gh actions-lock has not been shown to regenerate (this-ssg release.yml slsa-github-generator@f7dd8c54, run 43156768667)
SQ-LOCK-03 FIX reusable-calling workflow ends with zero jobs; the lock records the callee's old owner_id after a transfer Repoint uses: to the canonical owner/name at a standards commit whose allowlist already covers metadatastician/standards/.github/workflows/*@* (8d1a2aa or later; governance / Allowlist Preflight reads the allowlist at the pinned commit), and relock with scripts/relock-standards-pin.sh from standards main Each record's owner_id equals the repository API owner id; the next run has jobs; check-allowed-actions.sh <allowlist at target> .github/workflows reports 0 gaps The pinned SHA is not reachable from the callee's default branch, or the callee is outside the estate, or no standards commit with that allowlist pattern is also free of pin-contract violations for the repository
SQ-PIN-01 FIX a pinned reusable ends startup_failure on every push while a newer audited SHA runs green elsewhere Repin to the newest audited green SHA and update the closure rows and pinned-reusable-sources.json together The next push run has jobs No audited closure row exists for the target SHA, or the new SHA declares more secrets
SQ-PIN-02 FIX the DEED lane fails unrecognised #-form on a literal the current ABNF admits: the repository pins an old standards SHA Bump every standards pin site in .github/workflows and scripts to one current SHA No pin site keeps the old SHA and validate-deed.sh passes at the new pin The newer linter rejects existing deed content; deed meaning is never edited
SQ-PIN-03 FIX setup-bun fails on the GitHub API rate limit, or a setup-bun step has no exact bun-version Set bun-version to the estate pin 1.4.2 standards scripts/check-setup-bun-pinned.sh passes The repository pins another Bun on purpose, or the step is in a third-party, forked or vendored workflow
SQ-PIN-04 FIX cargo install --git of the old panic-attacker name with no pin in a workflow Use hyperpolymath/panic-attack at the pin in gitbot-fleet bots/panicbot/canon/pin.toml No unpinned --git install remains The name maps to more than one repository
SQ-PIN-05 FIX a workflow whose job is a cross-repository uses: <owner>/<repo>/.github/workflows/<x>.yml@<sha> ends failure or startup_failure with zero jobs on every push Replace the caller with the inline-job form that rsr-template-repo main uses for the same workflow (it runs the gates from the pinned SHA without a cross-repository workflow_call) and keep actions.lock in sync The next push run is success with more than 0 jobs and the lock-sync gate passes The template has no inline form for that workflow, a required context name would change, the repository customised the caller on purpose, or the inline form still gives 0 jobs
SQ-TOOL-01 FIX different nickel or yq versions across CI, Containerfile, mise and setup scripts One version everywhere, sha256-pinned (Nickel 1.18.0; yq v4.53.3) One version found; the installer verifies the sha256 No upstream checksums exist (state trust-on-first-download, never verified)
SQ-TOOL-02 FIX Dependabot fails with Dependabot detected that Bun is misconfigured ... This project has a bun.lock file but no package-lock.json, yarn.lock or pnpm-lock.yaml (type misconfigured_tooling, tool Bun): a dependabot.yml entry uses package-ecosystem npm for a directory that bun manages In .github/dependabot.yml change package-ecosystem from npm to bun for that directory only, and correct any comment that says npm supports bun.lock; leave every other entry and its schedule, limits, groups and ignores unchanged The next Dependabot run for that directory finishes without a misconfigured_tooling error The directory holds a second lockfile (bun.lock plus another), or the repository has no matching Dependabot ecosystem
SQ-GEN-01 FIX estate-rules reports the committed REPOSITORY-MAP.adoc is stale Run bash scripts/gen-repo-map.sh . and commit the map tests/shape/repo_map_determinism_test.sh passes The map still differs after regeneration
SQ-GEN-02 FIX a deed or k9 line is over 998 octets (K9-E006) Wrap the string or description, layout only deed_lint.js and k9-validate --strict pass A wrap would change a value (identifier, hash, URL)
SQ-GEN-03 FIX a job with if: always() and needs: that downloads an artifact (actions/download-artifact without run-id) runs after a cancelled run and fails on the missing artifact (Hypatia WF029) Change the job condition to ${{ !cancelled() }} The workflow is valid YAML, the job still runs after failed or skipped needs, and it does not run in a cancelled run The job must run after cancellation on purpose, or the workflow is vendored or third-party
SQ-GEN-04 FIX kyaml-format: <file> is not KYAML (quintessence run 43156792336: build/container/compose.yaml) Run scripts/kyaml-format.sh <file> and commit the layout change. The generator is named here and is never taken from the log; it needs mikefarah Go yq (Python yq corrupts the file) The output is KYAML and the repository kyaml check passes; a missing yq means the check did not run, never a pass The output is not KYAML, the formatter changes a value, or yq is not the mikefarah Go build
SQ-CANON-01 FIX an instance's canon-lockstep job is an old inline copy and goes red on a canon release Run rsr-template-repo scripts/adopt-canon-lockstep.sh The canon-lockstep job passes with the lag notice The script refuses (lock coverage), or the pin itself must change (A2ML is never edited automatically)
SQ-PERM-01 FIX Resource not accessible by integration at get-a-workflow-run in a job that uploads SARIF on a private repository Add actions: read to that job's permissions, with a comment that names the reason The upload step passes on a re-run The permission is already present, or the repository is public
SQ-RUN-01 ACT a setup or download step fails with API rate limit exceeded and the same workflow passes on an adjacent commit Re-run the failed jobs once after the rate-limit reset, only if that commit is still the head The re-run passes It fails twice on the same commit, the log has no rate-limit text, or the step is not a setup or download step
SQ-RUN-02 ACT a PR is blocked with required checks green under a strict ruleset and its base is behind the default branch Update the branch by merging the base; never rebase a branch the squabbler did not open The required checks pass on the new head Merge conflict, or still blocked on an up-to-date head
SQ-RUN-03 ACT a step fails on a transient host or API error: gh: HTTP 5xx, a git remote returning 5xx, Could not resolve host, or Failed to connect Re-run the failed job once, only if that commit is still the head The re-run passes It fails twice on the same commit, or the status is 401 or 403 (a token problem: SQ-OWN-09, or a rate limit: SQ-RUN-01)
SQ-RUN-03 ACT a step fails on a transient host or API error: gh: HTTP 5xx, gh: Server Error (HTTP 5xx), a git remote returning 5xx, Could not resolve host, or Failed to connect Re-run the failed job once, only if that commit is still the head The re-run passes It fails twice on the same commit, or the status is 401 or 403 (a token problem: SQ-OWN-09, or a rate limit: SQ-RUN-01)
SQ-RUN-04 ACT gitleaks fails with fatal: ambiguous argument <sha>^..<sha>: unknown revision or path not in the working tree, because a shallow checkout lacks the range base commit; trufflehog reports commits: 0 for the same range Re-run the failed job once, only if that commit is still the head The re-run passes and the scan covers more than 0 commits It fails twice on the same commit: refer with the suggestion fetch-depth: 0 on the scan checkout
SQ-CODE-01 FIX a Rust readiness or integration test cannot find its binary under target/debug/build Use PathBuf::from(env!("CARGO_BIN_EXE_<bin>")); a code fix, first-line only when the code-fix guard passes cargo test --test readiness passes The fix would hit the code-fix unsafe list, the code is vendored or third-party, or the check does not pass on the PR head
SQ-CODE-02 FIX Validate Hypatia Baseline fails on new code_safety findings Fix the code behind the finding; never touch .hypatia-baseline.json The baseline job passes with the baseline file unchanged The fix would hit the code-fix unsafe list, the code is vendored or third-party, or the check does not pass on the PR head
SQ-CODE-03 FIX Sonar S7688 or S7679 in shell scripts Make the mechanical edit; decline S1192 in test files, where repetition is a convention bash -n, shellcheck and the repository's shell tests pass The fix would hit the code-fix unsafe list, the code is vendored or third-party, or the check does not pass on the PR head
SQ-CODE-03 FIX Sonar S7688, S7679 or S131 in shell scripts Make the mechanical edit: for S7688 rewrite [ ] to [[ ]] only in files with a bash shebang (never /bin/sh or POSIX scripts); for S131 add a no-op default case *) ;;; decline S1192 in test files, where repetition is a convention bash -n, shellcheck and the repository's shell tests pass An S7688 test uses -a or -o or relies on word splitting, the fix would hit the code-fix unsafe list, the code is vendored or third-party, or the check does not pass on the PR head
SQ-REF-06 REFER the DEED job fails could not read Username for https://github.com on a private repository Root cause: OTP 28 (28.3 and 28.5.0.8) miscompiles Hypatia.Rules.PinSites.walk/4; the template #328 shape only shifts memory layout. The fix is hyperpolymath/hypatia#1062. Never change the workflow to hide it DEED syntax passes on a private PR Always, until hypatia#1062 lands
SQ-REF-08 REFER Repo self-tests fails in dismiss-code-scanning-alerts-test.sh with <file>:<line> is not a write permission after lines were added above a permissions: block that .github/code-scanning-dismissals.json keys by line number Suggest keeping the recorded line where it is: put new comments inline or below it; never re-key the dismissal (it must match the live Scorecard alert's start_line on main) The test reports 0 failed Always when the permission line itself must move or change (re-keying after Scorecard re-scans main is the owner's call)
SQ-CODE-04 FIX an adapter prints a valid findings array but exits other than 0 or 1 and is reported as ran Accept arrays only for exit 0 or 1 (fleet-report-back.sh translate_or_error) Tests for exits 2, 3, 9 and 137 The fix would hit the code-fix unsafe list, the code is vendored or third-party, or the check does not pass on the PR head
Expand All @@ -50,9 +52,13 @@ SQ-OWN-06 OWNER a bot PR shows only external checks and no Actions runs: the wor
SQ-OWN-07 OWNER analyze (actions, none) stays red while GitHub's default code-scanning setup is on Report: the owner turns default setup off; never work around it in a workflow (owner) (owner)
SQ-OWN-08 OWNER a workflow ends startup_failure with 0 jobs and no check-runs or annotations while the byte-identical file with the same SHA pins succeeds in rsr-template-repo Report: the cause is repository-level (Actions policy, allowed-actions list, Pages setting); the owner reads the failed run's banner or Settings > Actions; never rewrite the workflow to chase it (owner) (owner)
SQ-OWN-09 OWNER gh: Bad credentials (HTTP 401), or git returned error: 401, from a token the workflow uses Report: the token is invalid, expired or revoked and the owner rotates it. Never retry (a retry cannot succeed) and never swap in another credential. A 403 is not matched here: it can be a rate limit (SQ-RUN-01) or a missing permission (SQ-PERM-01) (owner) (owner)
SQ-OWN-10 OWNER actions/create-github-app-token fails with Private Key is in OpenSSH format, error:1E08010C:DECODER routines::unsupported, or Invalid keyData Report: the App private key secret holds a key in the wrong format; the owner generates a new private key for the App and saves the whole file, whose first line is -----BEGIN RSA PRIVATE KEY-----. Never put the key in chat (owner) (owner)
SQ-OWN-10 OWNER actions/create-github-app-token fails with Private Key is in OpenSSH format, error:1E08010C:DECODER routines::unsupported, or Invalid keyData Report: the App private key secret holds a key in the wrong format; the owner generates a new private key for the App and saves the whole file in the RSA PEM form that the App settings page produces (not an OpenSSH key). Never put the key in chat (owner) (owner)
SQ-OWN-11 OWNER a benchmark gate fails with a ratio over its limit (for example escape_hatches 2.053x, or 2.285x against a 2.0x limit) Report: a performance regression, or a limit that needs the owner judgement; never raise the limit (owner) (owner)
SQ-OWN-12 OWNER the same scanner (panic-attack) is pinned at more than one commit across the estate workflows Report the sites that differ from the ruled pin (owner ruling D7: panic-attack 410a8f82, pons bca1ea28); the change goes through the standards applier, never a hand edit (owner) (owner)
SQ-OWN-13 OWNER a workflow push is refused by a repository ruleset: gh: Repository rule violations found / Code coverage checks require a pull request. (HTTP 409) (Apodeixis.jl run 43257204071, Publish the lane badges, red on its base too) Report: a ruleset requires a pull request for that push, so the workflow cannot push to the protected branch. The owner changes the workflow to open a PR or adjusts the ruleset. The squabbler never changes a ruleset and never pushes around one (owner) (owner)
SQ-OWN-14 OWNER <file>: non-v8 UUID literal (<uuid>) / UUID check failed. Use ESTATE-UUID-V8 (ADR-008) (Apodeixis.jl run 43257204302, ./Project.toml) Report: an identifier needs the owner or the A7 identity process. The squabbler never mints, rewrites or retypes a UUID (owner) (owner)
SQ-OWN-15 OWNER SonarQube ERROR Fail to load the list of plugins installed on the server: Not authorized or organization not found, then sonar-scanner failed with exit code 3 (lithoglyph run 43152462261) Report: the Sonar token or organization is missing or wrong; the owner fixes the secret or the Sonar project. Never remove or skip the Sonar step (owner) (owner)
SQ-OWN-16 OWNER a workflow force- or mirror-pushes to a forge, gated on a *_MIRROR_ENABLED variable, with no check of the source repository visibility anywhere in the file (Hypatia WF030, found statically: standards mirror-reusable.yml gates the job at line 93 and runs git push --force gitlab main at line 144; github-backup-mirror-reusable.yml too) Report: the owner decides the guard (refuse, or skip loudly, when the source is private and the forge copy is not confirmed private; escalate when the forge visibility cannot be read). The squabbler never edits a mirror workflow, never pushes to a forge and never changes a visibility setting (owner) (owner)
SQ-NOI-01 NOISE Fork governance failed (deed=cancelled) on a head that a newer commit replaced Label superseded; read the newest head The newest head is not deed=cancelled deed=cancelled on the newest head
SQ-NOI-02 NOISE couldn't find remote ref refs/pull/<N>/merge and the PR closed before the job started None: do not re-run a closed PR The PR API shows the PR closed The PR is still open (conflict or deleted ref)
SQ-NOI-03 NOISE Inbox Steward ends startup_failure only on runs triggered by chatgpt-codex-connector[bot] None Count of startup failures by actor Inbox Steward becomes a required check
Loading
Loading