Content
70%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 thorough, well-sequenced extraction workflow with strong validation and concrete templates, undermined by heavy repetition of the prototype-screen rule and large inline reference/background sections that belong in separate files. Actionability is good but would be fully executable if MCP tool names were specified.
Suggestions
State the prototype-screen 3-task / subagent rule once (e.g. in step 7) and reference it from 'Critical Requirements', 'Common Mistakes', and 'Implementation Notes' instead of restating it verbatim in ~6 places.
Move the Background section (Content Types table, Board vs Item URLs, How Images Are Extracted, Best Practices, Troubleshooting) into references/ files and link from SKILL.md to reduce inline bulk and improve progressive disclosure.
Replace the repeated 'the appropriate Miro MCP item-retrieval/image tool' phrasing with the specific MCP tool names so the guidance is fully copy-paste executable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The core operational guidance (paths, JSON schemas, subagent prompt, curl) earns its tokens, but the prototype-screen 3-task rule is restated in ~6 places (step 5, step 7, 'Critical Requirements', Common Mistakes 2–6, Implementation Notes) and the inline Background/Troubleshooting sections add padding that could be tightened. | 3 / 5 |
Actionability | Concrete file paths, a copy-paste subagent prompt template, an executable curl download command, and explicit JSON structures make this mostly executable; the gap is the repeated placeholder 'the appropriate Miro MCP item-retrieval/image tool' instead of named tool calls. | 4 / 5 |
Workflow Clarity | A clearly sequenced 9-step workflow with an explicit verification checklist (step 9), a feedback loop ('If verification fails ... re-extract missing items'), and per-item task tracking; validation is present so the batch/destructive cap does not apply. | 5 / 5 |
Progressive Disclosure | Section structure is good and the one reference (references/spec-storage.md) is clearly signaled and exists, but ~100 lines of reference material (Background, Content Types table, Best Practices, Troubleshooting) are inlined in SKILL.md rather than split into referenced files. | 3 / 5 |
Total | 15 / 20 Passed |