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-structured pipeline skill: a clearly sequenced 7-phase workflow with explicit validation and feedback loops for batch extraction, lean prose, and concrete tooling references. The only notable improvements are an example recipe YAML and moving recipe details into a reference file.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and imperative with no explanations of concepts Claude already knows; the only trimming opportunities are the "Contract" section and opening paragraph, which partially restate the frontmatter description. | 4 / 5 |
Actionability | Guidance is concrete throughout — recipe file path ("~/.gbrain/recipes/{name}.yaml"), specific tool names (put_raw_data, file_upload), explicit dedup match rules, and a worked output-format example — but an example recipe YAML block would make the recipe scaffolding fully copy-paste ready. | 4 / 5 |
Workflow Clarity | The 7-phase sequence is explicit with validation checkpoints where they matter for batch operations: the numbered EXTRACTION INTEGRITY RULE (save raw first, re-read from saved files, never trust LLM working memory), the dedup gate before tracker updates, and a fail-improve loop that logs LLM fallbacks for regex improvement. | 5 / 5 |
Progressive Disclosure | Sections are well organized at overview altitude with a one-level reference to skills/conventions/quality.md, and no bundle files exist to misplace; the minor gap is that recipe schemas and the three built-in recipes are described inline and could be split into a reference file for a cleaner overview. | 4 / 5 |
Total | 17 / 20 Passed |