Compute a quick quality readiness signal for the current working tree.
Process
- Run
make check && make test-e2e. - Fail = 0: if either target fails, score is 0 (auto-fail per quality-gates: tests/build broken is CRITICAL).
- Pass = 100 baseline: if both pass, the static gate is green. For a full rubric-based score (Major/Minor deductions), invoke
/gemini-reviewor/advanced-reviewand subtract:- 10 per Major finding
- 3 per Minor finding
- Report against the three thresholds from
rules/quality-gates.md:- ≥80 commit-ready
- ≥90 PR-ready
- ≥95 excellence
Output format
Score: <n>/100
Gate: <commit-ready | PR-ready | excellence | BLOCKED>
Breakdown:
make check: <PASS|FAIL>
make test-e2e: <PASS|FAIL>
Major findings: <n> (-<10n>)
Minor findings: <n> (-<3n>)
Ready to commit: <yes|no>
Ready to open PR: <yes|no>
Excellence: <yes|no>
Rules
- Be honest. If the output is ambiguous, report "inconclusive" and say why. Do not inflate.
- CRITICAL trumps all. A failing
make check/test-e2eyields score 0 regardless of other factors. - Missing gate = not zero. If
make checkormake test-e2etargets don't exist in the project, do NOT claim score 100. Report "gate missing, run/project-checksto scaffold" and refuse to score. - Review integration is optional. If the user didn't ask for a full review, report the static gate result only and note that a rubric-based score requires a review agent pass.
When to use
- Before
git commit: quick readiness check. - Before opening a PR: confirm ≥90 threshold.
- Mid-session checkpoint: answer "are we done?" with a number instead of vibes.
When NOT to use
- Pre-commit already running as a hook: the hook gives a go/no-go;
/scoregives you a number and a breakdown. Different use. - Large refactors where full review is warranted: go straight to
/advanced-review, not/score.