Content
81%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 well-built instruction skill: fully actionable phases, a concrete output contract, and explicit error-recovery paths make it easy to execute correctly. The main cost is token redundancy — the same three boundary points (single-idea scope, read-only mode, concept-synthesis handoff) are each repeated four to five times across sections, which could be consolidated without losing clarity.
Suggestions
Consolidate the concept-synthesis contrast into one place (e.g., 'What this is not') and reference it from the Contract, Phase 5, Anti-Patterns, and Related Skills instead of restating it; the same applies to the read-only/no-mutation rule, which is currently stated in five sections.
Drop the 'Tools Used' section or fold its one-line tool descriptions into Phase 2 where the tools are already invoked, removing a full re-enumeration of the frontmatter tool list.
Trim 'What this solves', which restates the frontmatter description, down to just the canonical example queries plus the one-sentence scope distinction.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The core guidance (bucket taxonomy, output template, quality rules) is dense and non-obvious, but boundary points repeat systematically: the concept-synthesis contrast appears in 'What this is not', the Contract, Phase 5, Anti-Patterns, and Related Skills; the read-only rule is stated five times; and 'Tools Used' re-enumerates tools already covered in Phase 2. This is 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the minor-trimming of the score-4 anchor. | 3 / 5 |
Actionability | For an instruction-only skill the guidance is fully executable: a copy-paste-ready output template with placeholders and a concrete citation format ('[source-id:slug], "short quote"'), named tools per phase, a seven-bucket taxonomy with definitions, canonical example queries, and specific failure remedies. This matches the 'fully executable; specific examples cover the common cases' anchor. | 5 / 5 |
Workflow Clarity | Five clearly sequenced phases with explicit checkpoints: a disambiguation ask in Phase 1, gap-labeling ('No clear evidence found' plus what was checked) in Phase 3, and a dedicated 'When it fails' section with specific error codes (page_not_found, degraded-notice handling) and recovery steps. The skill is read-only, so the destructive/batch validation cap does not apply; validation and error-recovery loops are explicit rather than implicit. | 5 / 5 |
Progressive Disclosure | Sections are well-organized and external references (../conventions/quality.md, ../conventions/agent-operator-protocol.md) are one level deep and clearly signaled in an opening blockquote, fitting 'good structure; references mostly clear'. It falls short of the score-5 anchor because no bundle files exist to verify the referenced paths against, and sibling-skill and convention references cannot be confirmed as real files; the duplicated boundary sections also leave minor organization gaps. | 4 / 5 |
Total | 17 / 20 Passed |