Content
65%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 well-structured and mostly executable with concrete commands and a clear install flow, but it relies on reference files that are absent from the bundle and treats validation as optional before a destructive promotion step.
Suggestions
Ship the referenced reference/sop-structure-pipeline.md, reference/generator-worker-spec.md, and reference/operator-playbook.md in the bundle so the "Always"-required extraction and install details are actually loadable.
Promote validation from optional to an explicit checkpoint: run scripts/skill_gen/validator.py on the draft and only call skills.import_local after it passes, adding a fix-and-revalidate feedback loop.
Trim repeated restatements of the same reference paths across sections to reduce token cost and noise.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and action-oriented with no concept padding, but restates reference paths across several sections and uses a noisy bold-heavy inline style that could be trimmed. | 4 / 5 |
Actionability | Concrete CLI commands, function names, and paths are given, but the three "Always"-required reference/*.md files are cited yet missing from the bundle, leaving the extraction detail non-executable from the body alone. | 4 / 5 |
Workflow Clarity | A clear 4-step sequence with explicit promotion exists, but validation is marked optional rather than an enforced checkpoint for a destructive overwrite (force: true replacing an installed skill), so the destructive-cap rule limits this to 3. | 3 / 5 |
Progressive Disclosure | A "When to open reference files" table signals one-level-deep references and content is well split, but the referenced reference/*.md files do not exist in the bundle, breaking navigation against the actual structure. | 3 / 5 |
Total | 14 / 20 Passed |