You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Deferred from PR #1334: define NOT_SERVED_REPOS exclusions under standards/ and have scripts/compliance-audit.sh (check_required_workflows) and the deployer both consume it, so the audit stops filing findings for software-factory / software-factory-canary. #1334 (comment)
From PR #1334 (CodeAnt nitpick, issue comment IC_kwDORyesfc8AAAABbykAlQ): NOT_SERVED_REPOS in scripts/deploy-standard-workflows.sh excludes repos only from deployment; the compliance audit still checks every org repo for required workflows, so intentionally absent stubs in software-factory/software-factory-canary stay as permanent findings. Follow-up: teach the audit to honor the same list.
Deferred from PR #1196: #1196 (comment) — cubic P3: add a Bats regression case for the ruleset-bypass-unfetchable-<id> fail-closed branch in scripts/compliance-audit.sh (failed ruleset detail fetch → error finding, mark_repo_inconclusive, no closure of existing issues). Needs new scaffolding because tests/ has no harness that sources compliance-audit.sh without running main.
Deferred from PR #1428 (CodeRabbit summary comment IC_kwDORyesfc8AAAABgp2x9Q, Linked Issues check warning): #1410's concrete case — declare mutation (all) in empower-personal-dashboard/.github/required-checks.json, apply to code-quality, and remove the interim mutation-testing ruleset — lives in another repo and is the post-merge rollout step.
Deferred from PR #1505 (CodeRabbit summary comment IC_kwDORyesfc8AAAABk8Esgw): Docstring Coverage pre-merge warning (57.78% vs 80%) on a mechanical [[ ]] refactor (S7688, #1462). Adding docstrings to touched shell functions is unrelated to this change.
Deferred from PR #1499 (#1499 (comment)): backfill/verify count a matching job as present without checking its workflow's triggers (push-only to main, path filters), so a job that never runs on pull requests can be skipped or pass verify. Real but a larger design change (event/branch/path analysis); not part of the initial backfill script.
PR #1506 deferred finding: #1506 (comment) — codeant-ai notes _reusable_differs returns 0 when the destination ring has no tag (empty prior), so failures in _pair_state can be labelled PRE_EXISTING. Pre-existing logic untouched by #1506 (a bracket/octal refactor); needs its own behavioral change + test.
Deferred from PR #1499 (CodeRabbit pre-merge 'Linked Issues' check): the post-merge rollout for #581 — merge each backfill standards-sync PR, resolve NO_CI/CONFLICT repos, confirm backfill-ci-quality-jobs.sh --verify --all exits 0, then add coverage and Secret scan (gitleaks) to standards/rulesets/code-quality.json and run apply-rulesets.sh --all. Operational follow-up that cannot happen inside this PR.
PR #1499 — #1499 (comment) : backfill-ci-quality-jobs.sh does not validate that a repo's existing ci.yml has pull_request/merge_group triggers, so a push-only ci.yml can pass the backfill yet never report the new contexts on PRs. Follow-up design change.
From PR #1504 (CodeRabbit summary comment IC_kwDORyesfc8AAAABk8DxSQ, #1504): hardening proposal — apply the same typed/bounded total validation used in sonar_list_projects to the per-repo issue listing in scripts/sonarcloud-audit.sh (~line 191-198), failing before a repo is marked scanned.
PR #1505 (CodeRabbit summary comment IC_kwDORyesfc8AAAABk8Esgw): Docstring Coverage pre-merge warning (58.70% vs 80%) on touched shell functions. Out of scope for the behavior-preserving [[ ]] refactor; track separately. #1505
From PR #1117 (codeant-ai nitpick, comment IC_kwDORyesfc8AAAABUNLrVg): promote-all runs under the fleet concurrency group (.github/workflows/canary-rollout.yml:136) while rollback uses the per-agent group, so a concurrent promotion and rollback could race on the same channel tag. Pre-existing workflow design (#1025), outside the #1103 rollback-resolution fix.
Deferred from PR #1299: #1299 (comment) — probe-then-force-reset of the sync branch in sd_deploy_files_via_pr has a small race window with concurrent same-day runs; fix is a unique per-run branch name (caller change) or a lock.
Also deferred from PR #1299 (CodeRabbit summary IC_kwDORyesfc8AAAABa-q7jA): the force-reset of the sync branch does not record the discarded previous head SHA; consider logging it to stderr before the PATCH for recovery.
Deferred from PR #1299 (CodeRabbit summary comment IC_kwDORyesfc8AAAABa-q7jA, Security Architecture Review): (1) log/retain the previous head SHA before _sd_branch_reset force-resets a same-date sync branch in scripts/lib/standards-deploy.sh (needs a bats stderr-separation change); (2) the non-atomic check→reset window across overlapping same-day invocations — use unique per-run branch names or shared serialization.
PR #1299 (CodeRabbit summary comment IC_kwDORyesfc8AAAABa-q7jA, Security Architecture Review): force-reset of the date-based sync branch in sd_deploy_files_via_pr is not atomic with the no-open-PR probe, and the previous branch head is not saved for recovery. Mitigated by the scheduled workflow's concurrency lane; follow-up: unique per-run branch names or shared serialization across all callers, and log/retain the prior head SHA before reset.
Deferred from PR #1218 (): CodeRabbit Security Architecture Review hardening proposal — when the PR-description scorer that consumes sc_description_missing is introduced, treat nonzero helper status as uncertainty requiring escalation, and keep description completeness separate from independent authorization controls. Future integration safeguard; no current caller.
From PR #1245 (CodeRabbit summary comment IC_kwDORyesfc8AAAABZBWa6Q, Linked Issues check): record the CLI-resolved model ID and the claude-code-action version (currently v1.0.268) for --model opus, via a minimal/dry-run resolution check. Needs a live CLI run with credentials; cannot be produced in the review-fix pass. Code is a --model passthrough with no ANTHROPIC_MODEL override (covered by test/workflows/feature-ideation/model-input.bats).
PR #1299 (CodeRabbit summary comment IC_kwDORyesfc8AAAABa-q7jA, #1299): standards-deploy.sh branch force-reset is not atomic with the open-PR check, and the discarded head is not recorded. Fix needs per-run unique branches or shared serialization, plus a recovery log that does not break the stdout/stderr contract the bats suite asserts.
Activity
don-petry commentedon Oct 14, 2026
Deferred from PR #1334: define NOT_SERVED_REPOS exclusions under standards/ and have scripts/compliance-audit.sh (check_required_workflows) and the deployer both consume it, so the audit stops filing findings for software-factory / software-factory-canary. #1334 (comment)
don-petry commentedon Oct 14, 2026
From PR #1334 (CodeAnt nitpick, issue comment IC_kwDORyesfc8AAAABbykAlQ): NOT_SERVED_REPOS in scripts/deploy-standard-workflows.sh excludes repos only from deployment; the compliance audit still checks every org repo for required workflows, so intentionally absent stubs in software-factory/software-factory-canary stay as permanent findings. Follow-up: teach the audit to honor the same list.
don-petry commentedon Oct 14, 2026
Deferred from PR #1196: #1196 (comment) — cubic P3: add a Bats regression case for the
ruleset-bypass-unfetchable-<id>fail-closed branch inscripts/compliance-audit.sh(failed ruleset detail fetch → error finding,mark_repo_inconclusive, no closure of existing issues). Needs new scaffolding because tests/ has no harness that sources compliance-audit.sh without running main.don-petry commentedon Oct 29, 2026
Deferred from PR #1428 (CodeRabbit summary comment IC_kwDORyesfc8AAAABgp2x9Q, Linked Issues check warning): #1410's concrete case — declare
mutation (all)inempower-personal-dashboard/.github/required-checks.json, apply tocode-quality, and remove the interimmutation-testingruleset — lives in another repo and is the post-merge rollout step.don-petry commentedon Nov 9, 2026
PR #1299 thread #1299 (comment) — concurrent same-day runs can force-reset each other's sync branch (needs per-run branch or per-repo/branch serialization).
don-petry commentedon Nov 9, 2026
PR #1299 thread #1299 (comment) — concurrent same-day runs can force-reset each other's sync branch (needs per-run branch or per-repo/branch serialization).
don-petry commentedon Nov 10, 2026
Deferred from PR #1505 (CodeRabbit summary comment IC_kwDORyesfc8AAAABk8Esgw): Docstring Coverage pre-merge warning (57.78% vs 80%) on a mechanical
[[ ]]refactor (S7688, #1462). Adding docstrings to touched shell functions is unrelated to this change.don-petry commentedon Nov 10, 2026
Deferred from PR #1499 (#1499 (comment)): backfill/verify count a matching job as present without checking its workflow's triggers (push-only to main, path filters), so a job that never runs on pull requests can be skipped or pass verify. Real but a larger design change (event/branch/path analysis); not part of the initial backfill script.
don-petry commentedon Nov 11, 2026
PR #1506 deferred finding: #1506 (comment) — codeant-ai notes
_reusable_differsreturns 0 when the destination ring has no tag (emptyprior), so failures in _pair_state can be labelled PRE_EXISTING. Pre-existing logic untouched by #1506 (a bracket/octal refactor); needs its own behavioral change + test.don-petry commentedon Nov 11, 2026
Deferred from PR #1499 (CodeRabbit pre-merge 'Linked Issues' check): the post-merge rollout for #581 — merge each backfill standards-sync PR, resolve NO_CI/CONFLICT repos, confirm
backfill-ci-quality-jobs.sh --verify --allexits 0, then addcoverageandSecret scan (gitleaks)tostandards/rulesets/code-quality.jsonand runapply-rulesets.sh --all. Operational follow-up that cannot happen inside this PR.don-petry commentedon Nov 11, 2026
PR #1499 — #1499 (comment) :
backfill-ci-quality-jobs.shdoes not validate that a repo's existingci.ymlhaspull_request/merge_grouptriggers, so a push-only ci.yml can pass the backfill yet never report the new contexts on PRs. Follow-up design change.don-petry commentedon Nov 11, 2026
PR #1437 deferred findings:
don-petry commentedon Nov 11, 2026
#1505 (comment) — Normalize
issue_limitto base 10 in collect-signals.sh:113 to prevent octal parsing errors with zero-padded valuesdon-petry commentedon Nov 11, 2026
From PR #1504 (CodeRabbit summary comment IC_kwDORyesfc8AAAABk8DxSQ, #1504): hardening proposal — apply the same typed/bounded
totalvalidation used insonar_list_projectsto the per-repo issue listing inscripts/sonarcloud-audit.sh(~line 191-198), failing before a repo is marked scanned.don-petry commentedon Nov 11, 2026
PR #1505 (CodeRabbit summary comment IC_kwDORyesfc8AAAABk8Esgw): Docstring Coverage pre-merge warning (58.70% vs 80%) on touched shell functions. Out of scope for the behavior-preserving
[[ ]]refactor; track separately. #1505don-petry commentedon Nov 12, 2026
Deferred from PR #1218: #1218 (comment) — safety-checks.sh
sc_description_missingonly recognizes ATX headings; Setext headings (Problem+=====) don't open a section.don-petry commentedon Nov 12, 2026
From PR #1117 (codeant-ai nitpick, comment IC_kwDORyesfc8AAAABUNLrVg):
promote-allruns under thefleetconcurrency group (.github/workflows/canary-rollout.yml:136) while rollback uses the per-agent group, so a concurrent promotion and rollback could race on the same channel tag. Pre-existing workflow design (#1025), outside the #1103 rollback-resolution fix.don-petry commentedon Nov 12, 2026
Deferred from PR #1299: #1299 (comment) — probe-then-force-reset of the sync branch in
sd_deploy_files_via_prhas a small race window with concurrent same-day runs; fix is a unique per-run branch name (caller change) or a lock.don-petry commentedon Nov 12, 2026
Also deferred from PR #1299 (CodeRabbit summary IC_kwDORyesfc8AAAABa-q7jA): the force-reset of the sync branch does not record the discarded previous head SHA; consider logging it to stderr before the PATCH for recovery.
don-petry commentedon Nov 12, 2026
Deferred from PR #1299 (CodeRabbit summary comment IC_kwDORyesfc8AAAABa-q7jA, Security Architecture Review): (1) log/retain the previous head SHA before
_sd_branch_resetforce-resets a same-date sync branch inscripts/lib/standards-deploy.sh(needs a bats stderr-separation change); (2) the non-atomic check→reset window across overlapping same-day invocations — use unique per-run branch names or shared serialization.don-petry commentedon Nov 13, 2026
PR #1299 (CodeRabbit summary comment IC_kwDORyesfc8AAAABa-q7jA, Security Architecture Review): force-reset of the date-based sync branch in
sd_deploy_files_via_pris not atomic with the no-open-PR probe, and the previous branch head is not saved for recovery. Mitigated by the scheduled workflow's concurrency lane; follow-up: unique per-run branch names or shared serialization across all callers, and log/retain the prior head SHA before reset.don-petry commentedon Nov 13, 2026
Deferred from PR #1218 (): CodeRabbit Security Architecture Review hardening proposal — when the PR-description scorer that consumes
sc_description_missingis introduced, treat nonzero helper status as uncertainty requiring escalation, and keep description completeness separate from independent authorization controls. Future integration safeguard; no current caller.don-petry commentedon Nov 13, 2026
Link for the PR #1218 deferral above: #1218 (comment)
don-petry commentedon Nov 13, 2026
From PR #1245 (CodeRabbit summary comment IC_kwDORyesfc8AAAABZBWa6Q, Linked Issues check): record the CLI-resolved model ID and the claude-code-action version (currently v1.0.268) for
--model opus, via a minimal/dry-run resolution check. Needs a live CLI run with credentials; cannot be produced in the review-fix pass. Code is a--modelpassthrough with no ANTHROPIC_MODEL override (covered by test/workflows/feature-ideation/model-input.bats).don-petry commentedon Nov 15, 2026
PR #1299 (CodeRabbit summary comment IC_kwDORyesfc8AAAABa-q7jA, #1299): standards-deploy.sh branch force-reset is not atomic with the open-PR check, and the discarded head is not recorded. Fix needs per-run unique branches or shared serialization, plus a recovery log that does not break the stdout/stderr contract the bats suite asserts.