Running CI Preflight Agent Skill

Running ci:preflight

hogli ci:preflight scopes a curated set of checks to the files your branch touched — each mapped to a

CI failure class that has taken master down — plus an always-on branch-freshness check. It is the

pre-push counterpart to hogli ci:insights (what is _already_ broken on master).

The pre-push hook runs ci:preflight --strict automatically and blocks the push on failed checks.

Never bypass it with --no-verify — fix what it reports instead.

The loop (when the hook blocks, or before reporting done)

hogli ci:preflight --fix

• Run with --fix — it formats, lints, and auto-fixes what is safe.
• Read each line: ✓ pass, ✗ fail, → advisory (do it yourself), · skipped (capability absent).
• Resolve every ✗ fail — these are what --fix could not (real lint error, broken lockfile, migration conflict). These block the push.
• Act on every → advisory — e.g. openapi advisory → run hogli build:openapi and commit the drift; staleness advisory → git merge origin/master. Advisories never block, but ignoring them ships the failure to CI. Resolve them before pushing, including drift you didn't introduce — you own the branch state you push.
• Re-run until clean, then push.

Notes

• Strict = failures only. --strict (what the hook runs) exits non-zero only on ✗ fail — advisories are unverifiable-locally classes, so they warn without blocking. A clean exit means "nothing left to fix", not "CI will pass" — CI stays the authoritative gate. Non-blocking is a mechanism limit, not permission to skip.
• type-check is a nudge, not a run. Only a repo-wide mypy run matches CI (mypy follows imports, so a changed-file subset misses reverse-dependency breakage), and that costs minutes cold — too much to tax every push with. So preflight names the command instead of running it: judge whether your change is type-risky and run it yourself. CI blocks on the same command, so a type error you skip here comes back as a full re-run.
• Staleness is risk-based. It fires when merging master _now_ would actually break something — textual merge conflicts (computed via git merge-tree, working tree untouched), migrations added on both sides, generated-file inputs changed on both sides, or CI workflows changed on master — plus a behind/age backstop, aggressive by default (5 commits / 2 days; env-tunable via HOGLI_PREFLIGHT_STALE_COMMITS/HOGLI_PREFLIGHT_STALE_DAYS) so we over-warn to start and tune down from telemetry. Merge master in when it fires. Advisory only, never auto-merged.
• · skipped (needs stack/node) is expected on a bare checkout or sandbox. Start the stack with hogli start to run those, or let CI cover them. No hooks in your environment (no node_modules)? Run the loop yourself before pushing.
• Flags. --against <ref> diffs against an explicit base; --json emits a machine-readable summary.
• Kill switch. HOGLI_PREFLIGHT_DISABLED=1 makes the command (and the hook) a no-op with exit 0. It is a rollout/emergency lever — respect it; never unset it to force a run.

Why it matters

Drafts already run a trimmed CI subset; the expensive waste is a ready PR that fails the full matrix on

something deterministic, gets fixed, and re-runs the whole matrix. Catching that locally is the cheapest

CI saving available.