ci-triage

Shared logic for classifying a failed CI workflow run and recording the outcome. Used by humans (or Claude in an interactive session) when triaging a red main workflow or a red check on an open PR — the latter via the global pr-lifecycle skill, whose CI-triage section delegates the taxonomy here.

This skill owns the taxonomy, the de-dup rules for the main-red issue, and the issue template. Repo-specific reproduction steps and failure patterns live in the consumer repo's CLAUDE.md / <repo>-dev-process (read by pr-lifecycle), and e2e classification is delegated to web-testing.

Taxonomy

Every failure lands in exactly one bucket. Be explicit — "unclear" is not a bucket, needs-human is.

Bucket	Signal	Default action
`regression`	Reproduces deterministically on `main` at HEAD	File/update `main-red`, @-mention suspect author
`flake`	Same workflow passed on the previous main commit without code change	Comment on the existing `main-red` issue if one is open; otherwise skip
`infra`	Postgres service didn't come up, rustup 403, pnpm registry down	File/update `main-red` labelled `infra`; do not @-mention authors
`needs-human`	Logs truncated, auth failed, classification genuinely ambiguous	Open a `main-red` issue with raw log excerpt, label `needs-human`

Do not invent a fifth bucket. If the signal you see fits nothing above, it's needs-human.