Content
71%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is an exceptionally lean, well-sequenced workflow with a concrete command, explicit polling and a done-when checkpoint, and a precise report format. Its main weakness is that every command and GraphQL query is delegated to ./queries.md, which is missing from the skill bundle, leaving the executable layer and the progressive-disclosure path broken; the ../../guidelines/pull-requests.md links also point outside the bundle and are unverifiable.
Suggestions
Ship queries.md alongside SKILL.md (or inline the GraphQL queries and gh commands): the body delegates all commands for steps 2–3 to ./queries.md, which does not exist in the bundle, so those steps cannot currently be executed.
Add an explicit timeout or fallback for the polling loop (e.g., what to do if a check stays pending indefinitely or Greptile never posts a review) to provide an error-recovery feedback loop.
Verify the out-of-bundle links (../../guidelines/pull-requests.md, ../handle-pr-comments/SKILL.md) resolve where the skill is installed, or inline the one-line ci:* label rule they are cited for.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~30-line body is lean and efficient with zero padding: no concept explanations (it assumes Claude knows gh and GraphQL), and every line instructs ("Poll `statusCheckRollup` every 30 seconds", "Put human comments before bot comments"). The one explanatory clause — "because only GraphQL gives a thread's id and whether it is resolved" — is non-obvious domain knowledge that earns its place, so there is nothing to trim (not 4). | 5 / 5 |
Actionability | There is one copy-paste command (`gh pr view --json number -q .number`) and precise field names, intervals, and output format, but the actual commands and GraphQL queries for steps 2–3 are delegated to "[queries.md](./queries.md) has the commands and GraphQL queries for the steps below" — and that file is absent from the bundle (no references/, scripts/, or assets/ directories; queries.md does not exist), so key executable details are missing. Not 4 because the gap is central rather than minor; not 2 because what is written is specific and directive rather than high-level hints. | 3 / 5 |
Workflow Clarity | A clear numbered 4-step sequence with an explicit polling checkpoint ("Poll `statusCheckRollup` every 30 seconds until no check is pending or in progress") and an explicit completion condition ("Done when all checks are finished and each unresolved thread is in the table"). Not 5 because there is no error-recovery guidance (e.g., a check stuck pending, or Greptile never posting a review); not 3 because checkpoints are explicit, not implicit. | 4 / 5 |
Progressive Disclosure | The split is structurally right — a short overview delegating command/query detail to a clearly signaled, one-level-deep reference — but the referenced ./queries.md does not exist in the bundle, so the primary disclosure path is broken and the deferred content is unavailable. Not 4 because a missing main reference is more than a minor organization gap; not 2 because the body itself is well-structured and the reference is prominently signaled, not buried or inlined. | 3 / 5 |
Total | 15 / 20 Passed |