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
{{ message }}
Repository navigation
sdlc ci runs no lint or format check, so a --no-verify commit lands unlinted under a green required check #48
resolve(cfg, "ci") builds ci from the push tier minus CI_EXCLUDES, and push carries no ruff, so ruff-check and ruff-format exist only in the commit tier. The commit tier is the easiest one to skip (git commit --no-verify, or a seat that never ran pre-commit install), and nothing downstream catches it: the required sdlc ci check goes green on a tree with ruff violations.
Repos that move onto python-sdlc lose a CI lint gate they had. Found reviewing crunchtools/gatehouse#134, whose old ci.yml ran uv run ruff check src tests on every PR.
Adding ruff-check to a repo's [tool.sdlc.ci] checks doesn't work as a fix, because it runs ruff check --fix .. It rewrites the tree and drops lanes, and fixable violations pass instead of failing.
Proposed: a non-mutating lint verdict in the push tier (so ci inherits it): ruff check --no-fix . plus ruff format --check .. It's read-only, so it can share the main lane.
Ashigaru triage: feature (70%), needs a maintainer's decision
The issue is accurate. In sdlc, DEFAULTS("push") has no ruff check, and ruff exists only as the mutating ruff-check (--fix) and ruff-format checks in the commit tier (t_ruff_check and t_ruff_format, around lines 858-867). ci is the push tier minus CI_EXCLUDES, so sdlc ci never lints. The proposed fix adds a read-only lint builtin that runs ruff check --no-fix . and then ruff format --check .. It would be registered in BUILTINS and added to DEFAULTS("push"), and it must stay out of WRITES_TREE so it shares the main lane in lane_of. The tests in tests/test_plan.py that enumerate the push verdicts, and the CHANGELOG and docs, would need updating. Adding a default verdict changes what every repo's push and ci run, and could newly fail repos that currently pass. Decision for the maintainer: should lint be a new push-tier default that ci inherits, as proposed, or opt-in? If it is a default, should it also be added to the commit tier or stay push-only?
Answer the question above in a comment, then add ready-to-code. Ashigaru reads the answer first.
Activity
github-actions commentedon Nov 17, 2026
Ashigaru triage: feature (70%), needs a maintainer's decision
The issue is accurate. In
sdlc,DEFAULTS("push")has no ruff check, and ruff exists only as the mutatingruff-check(--fix) andruff-formatchecks in the commit tier (t_ruff_checkandt_ruff_format, around lines 858-867).ciis the push tier minusCI_EXCLUDES, sosdlc cinever lints. The proposed fix adds a read-onlylintbuiltin that runsruff check --no-fix .and thenruff format --check .. It would be registered inBUILTINSand added toDEFAULTS("push"), and it must stay out ofWRITES_TREEso it shares the main lane inlane_of. The tests intests/test_plan.pythat enumerate the push verdicts, and the CHANGELOG and docs, would need updating. Adding a default verdict changes what every repo's push and ci run, and could newly fail repos that currently pass. Decision for the maintainer: shouldlintbe a new push-tier default that ci inherits, as proposed, or opt-in? If it is a default, should it also be added to the commit tier or stay push-only?Answer the question above in a comment, then add
ready-to-code. Ashigaru reads the answer first.fix: sdlc ci runs a non-mutating ruff verdict (#48) (#49)
chore: python-sdlc 0.7.0 (#52)