Content
88%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 highly actionable, well-sequenced operational skill with real commands, explicit safety gates, and verification loops for destructive/batch operations. Its weaknesses are mild: some rule repetition across sections and a monolithic ~390-line body where the embedded prompt templates and snapshot schema could live in reference files.
Suggestions
Consolidate the apply-consent rule ('--apply is consent, do not wait for a second yes') into a single authoritative location (e.g., Safety rule 9) and reference it from Modes and §5 instead of restating it.
Move the per-PR agent prompt template and the apply-mode child prompt into files under references/ (e.g., references/agent-prompt.md) and point to them, trimming SKILL.md toward an overview.
Consider moving the full snapshot.json example into a reference file and keeping only the path and key fields (headSha, sweptAt, security, mergeStateStatus) inline.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes Claude's competence — real commands and JSON schemas with almost no explanation of known concepts — but the apply-consent rule is stated three times (Modes, Safety rule 9, §5 intro) and push-verification steps are repeated in Safety rule 11, §5b, and the apply child prompt, which could be consolidated. | 4 / 5 |
Actionability | Fully executable throughout: copy-paste-ready gh/git commands, complete JSON return schemas for subagents, and fully embedded per-PR and apply-mode child prompt templates, covering the common cases (forks, 403s, conflicts). | 5 / 5 |
Workflow Clarity | The procedure is clearly sequenced (resolve repo → list → fan-out → report → apply → final report) with explicit validation checkpoints for risky batch mutations: security gate before any mutation, BEFORE/AFTER head-SHA push verification, rebase abort rules, snapshot diffing, and explicit failure handling (push-403, blocked-conflicts). | 5 / 5 |
Progressive Disclosure | Good structure with a clear, one-level-deep reference ('Security heuristics: [references/security-checklist.md]', which exists and is also pointed to from the agent instructions), but at ~390 lines the body inlines material — the snapshot JSON example, per-PR agent prompt template, and apply child prompt — that could be split into reference files to keep SKILL.md a leaner overview. | 4 / 5 |
Total | 18 / 20 Passed |