Content
86%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.
A tight, well-structured instruction-only skill: every step names the exact tool or API surface to use, the loop is interactive with per-comment user confirmation, and there is zero padding. Its only real gaps are the absence of ready-to-run GraphQL query/mutation text and any guidance for failure cases like outdated threads or failed mutations.
Suggestions
Include the actual GraphQL query for reviewThreads and the resolveReviewThread mutation (or a short example of each) so the fetch and resolve steps are copy-paste ready rather than assembled from field names.
Add brief handling for edge cases such as isOutdated threads (which the fetch step requests but the workflow never uses) and failed mutations, to add the error-recovery feedback loop workflow_clarity's top anchor looks for.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~25-line body is lean and assumes Claude's competence: no explanation of what PRs or review comments are, and even the one explanatory parenthetical ("Use GraphQL, not REST: only it exposes thread ids and resolution state") is non-obvious domain knowledge that earns its tokens. This matches anchor 5 ('lean and efficient; every token earns its place'); nothing matches anchor 4's 'minor instances of over-explanation'. | 5 / 5 |
Actionability | Concrete commands and API specifics are given throughout: `gh pr view`, the exact GraphQL fields ("id, isResolved, isOutdated, path, line, and comments"), the `resolveReviewThread` mutation name, and enumerated AskQuestion options ("apply suggested fix / apply a different fix / reply only / skip"). However, no copy-paste-ready GraphQL query or mutation text is included, leaving a minor gap that keeps it at anchor 4 ('mostly executable guidance... with minor gaps') rather than anchor 5's fully copy-paste-ready standard. | 4 / 5 |
Workflow Clarity | The four-step sequence is clear with an early checkpoint ("Stop if no PR exists") and a per-comment interactive loop that asks the user before applying changes and before committing — genuine validation checkpoints that satisfy the batch-operation concern. It falls short of anchor 5 only by lacking explicit error-recovery/feedback loops (e.g., what to do when a thread is outdated, or a mutation fails), while clearly exceeding anchor 3's 'checkpoints missing or implicit'. | 4 / 5 |
Progressive Disclosure | This is an under-50-line single-task skill with no bundle files (references/, scripts/, assets/ are absent), and nothing in it needs external material. Per the rubric's simple-skill guidance, well-organized sections alone qualify for anchor 5: it has a clear title, a four-step overview, and a separate Notes section, with no inlined content that belongs in another file and no nested references. | 5 / 5 |
Total | 18 / 20 Passed |