Content
90%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 lean, executable, and well-structured with clear sections and signaled file references. The main improvement is converting the workflow guidance into an explicit numbered sequence with validation feedback loops.
Suggestions
Render the 'Workflow guidance' section as a numbered sequence (1. dense map → 2. typed/--members map → 3. verify findings → 4. act) with an explicit validate-and-retry loop so the feedback path is unambiguous.
Confirm the referenced bundle files (map.ts, selftest.mjs) actually exist under the skill dir, since no references/scripts/assets directories are present to validate them against.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense, operational prose that assumes Claude's competence — no basic-concept padding — and every section earns its tokens (e.g. Swift parser limitations are non-obvious tradeoffs, not common knowledge). | 5 / 5 |
Actionability | Provides a copy-paste-ready invocation ('node <this-skill-dir>/map.ts <repoRoot> [flags]') plus concrete flag enums and per-mode output shapes covering the common cases. | 5 / 5 |
Workflow Clarity | The two-tier workflow (dense recon then typed exports/--members for the decision) is sequenced with explicit verification gates ('verify every dense-tier finding...', 'spot-verify... with grep before acting'), but presented as guidance bullets rather than a strict numbered checklist with feedback loops. | 4 / 5 |
Progressive Disclosure | Well-organized into clearly headed sections with one-level-deep references to bundle files ('Backed by map.ts next to this file', 'node selftest.mjs (in this skill dir)'), though no references/scripts/assets bundle directories exist and the long Flags and Swift-specifics sections are dense inline content. | 4 / 5 |
Total | 18 / 20 Passed |