Content
63%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 a well-organized instruction skill with a clear workflow, strong evidence-grading guidance, and one genuinely executable code example, but it is held back by a duplicated inline template, a pseudocode update pattern, and a references section pointing at files that are missing from the bundle.
Suggestions
Include the five referenced bundle files (REPORT_TEMPLATE.md, RESEARCH_PROTOCOL.md, tool_usage_details.md, TOOLS_REFERENCE.md, EXAMPLES.md) in the skill bundle — currently the links point at files that do not exist — and place them under references/ with matching relative paths.
Remove the ~70-line inline report template (or reduce it to a skeleton) since it duplicates REPORT_TEMPLATE.md, keeping the detailed version only in the referenced file.
Replace the comment-only "Progressive Update Pattern" code block with an actual executable snippet (read file, replace section, write back) or move it into RESEARCH_PROTOCOL.md with concrete code.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Dense, useful tables (dimensions, evidence grading, conflict handling), but the ~70-line report template is inlined even though REPORT_TEMPLATE.md is supposedly available, the "Disease Mechanism Reasoning" section spells out a causal cascade Claude can reason through itself, and the "Progressive Update Pattern" block is four comment lines. Mostly efficient but could be tightened — matching the anchor-3 example's profile. | 3 / 5 |
Actionability | Fully executable ZOOMA examples with real parameters and expected outputs, a concrete cross-skill command, and explicit per-context citation formats. Not 5 because most dimension tool calls are only named (deferred to tool_usage_details.md) and the core "Progressive Update Pattern" is comment-only pseudocode. | 4 / 5 |
Workflow Clarity | Clear create-research-update-cite sequence plus an end-of-run quality checklist that verifies sections, citations, and leftover placeholders. Not 5 because there are no explicit per-step validation or error-recovery loops (e.g., what to do when a tool call fails or returns nothing). | 4 / 5 |
Progressive Disclosure | The "Reference Files" section is well signaled and one level deep, but none of the five referenced files (REPORT_TEMPLATE.md, RESEARCH_PROTOCOL.md, tool_usage_details.md, TOOLS_REFERENCE.md, EXAMPLES.md) exist in the bundle, and the report template that should live in a separate file is inlined instead. Structure and signaling are present but the actual bundle does not support the claimed structure. | 3 / 5 |
Total | 14 / 20 Passed |