Content
62%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 content is well-structured and actionable, giving Claude concrete files, parsing rules, and output formats with clear pre-flight validation. Its main weakness is progressive disclosure: bulky output templates and calculation logic are inlined rather than split into bundled reference files.
Suggestions
Move the full status and single-track output templates into a bundled file (e.g. references/output-templates.md) and reference it from SKILL.md to reduce token cost.
Add an explicit retry/feedback step for malformed plan.md or metadata.json parsing rather than only erroring on missing files.
Tighten the calculation-logic section by linking to a scripts helper for task counting and progress-bar rendering if a scripts bundle is appropriate.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient with concrete file paths and output templates, but the large fixed-width status output blocks (full and single-track) are long and could be tightened or moved to a bundled reference file. | 3 / 5 |
Actionability | Guidance is mostly executable: it specifies exact files to read (conductor/product.md, tracks/{trackId}/plan.md), regex patterns for task counting, and progress-bar formulas with only minor gaps around how metadata.json and spec.md are parsed. | 4 / 5 |
Workflow Clarity | A clear sequenced workflow (pre-flight checks -> data collection -> output) with validation at init, and error states for missing files and unknown tracks; minor gaps since there is no explicit retry/feedback loop for malformed parses. | 4 / 5 |
Progressive Disclosure | Sections are organized but the sizable output templates and calculation logic are all inlined in SKILL.md with no references to separate files, and there are no bundle files to offload detail to. | 3 / 5 |
Total | 14 / 20 Passed |