Conciseness

Nearly every sentence carries repo-specific, non-derivable detail ("45k tracked files and 4.6 GiB of packed objects", "500 runs per 10s per repo", WF rule IDs, measured run links) and nothing explains concepts Claude already knows, placing it above the 3 anchor; but it is not the 5 anchor's "every token earns its place" — the ~17-line failure-rate-metric passage ("joins `github_workflow_jobs` to that table on `run_id`, then scopes the job side to a single `run_attempt`") and some restated draft/ready rules exceed what the authoring decision needs.

Actionability

Copy-paste-ready YAML for the common cases (canonical concurrency block, per-SHA group expression, `create-github-app-token` step with `|| github.token` fork fallback, sparse-checkout recipe, `curl -fsSL --retry 5 --retry-all-errors`, the `set -f` argument-splitting fix) plus exact commands (`hogli ci:plan`, `bin/hogli lint:workflows`) and named canonical files to copy from ("`ci-paths-filter.yml` is the smallest complete example") — the 5 anchor, not the 4 anchor, since the placeholders (`@<sha> # v6`) are deliberate templates backed by pointers to real files.

Workflow Clarity

A clear sequence with explicit validation checkpoints and feedback loops: check "things already tried" first, copy a canonical file, then validate with `bin/hogli lint:workflows` and `actionlint` before pushing, check conditions with `hogli ci:plan`, and "update its expectation and run the suite" (`hogli test:workflows`) when a condition changes — closing with a 13-item "New-workflow checklist". Error-recovery paths are stated (linter rejections take `# hogli-lint: allow-* -- <reason>`), matching the 5 anchor's checklists and feedback loops; no destructive/batch cap applies since this is authoring guidance.

Progressive Disclosure

Section structure is clear and the external pointers that exist are one-level-deep and well signaled ("Read its [ordered-image reference](../gating-production-deploys/references/ordered-image-releases.md)", "See [`/depot-ci`](../depot-ci/references/posthog-check-run-semantics.md)"), but the skill ships no references/ files of its own and inlines ~550 lines, including multi-screen deep dives (gate conclusion semantics, the warehouse failure-metric rules, the checkout duration table, the Depot consumers list). That matches the 3 anchor's "content that should be separate is inline" — above the 2 anchor because structure and signaling are good, below the 4 anchor because the inline deep-dives are more than "minor organization gaps".

Specificity

It names the domain ("adding or editing a GitHub Actions workflow, composite action, or reusable workflow under `.github/`") and enumerates a comprehensive set of concrete tasks — "new CI jobs, triggers, matrices, checkout/clone tuning, action pinning, GitHub App token auth, concurrency groups, `timeout-minutes`, `paths` filters, caching, or runner choice" — with no gaps inside its stated scope, matching the 5 anchor's "multiple specific concrete actions; comprehensive coverage" rather than the 4 anchor's "minor gaps".

Completeness

Both questions are answered explicitly: what ("Covers PostHog's workflow-authoring conventions and the reasons behind them... Points to the linters (`bin/hogli lint:workflows`, actionlint)") and when ("Use when adding or editing a GitHub Actions workflow... under `.github/`") with concrete trigger phrases, plus negative scope ("Not for debugging red CI (use debugging-ci-failures) or wiring a new secret end to end"). This is the 5 anchor verbatim in structure, clearly above the 4 anchor's "'when' could be more explicit".

Trigger Term Quality

Natural terms a user would actually say are all present — "GitHub Actions workflow", "CI", "composite action", "reusable workflow", "caching", "runner choice", "concurrency", `timeout-minutes`, `paths` filters — including the umbrella synonyms (GitHub Actions / CI) and the path anchor `.github/`, matching the 5 anchor's comprehensive synonym coverage rather than the 4 anchor's "a few natural terms missing".

Distinctiveness Conflict Risk

A clear niche — authoring `.github/` workflows at PostHog — with active disambiguation from siblings ("Not for debugging red CI (use debugging-ci-failures) or wiring a new secret end to end (use managing-github-actions-secrets)"; "Points to... the narrower skills for production deploys, secrets, and Depot runners"), so the wrong-skill trigger risk is minimal, matching the 5 anchor rather than the 4 anchor's "minor overlap risk".