Content
85%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is concise and clearly sequenced with a built-in source-tracing checkpoint, but it would be more actionable with a concrete example of the findings-file format and per-claim source citation.
Suggestions
Include a short example Markdown snippet showing one finding with its source citation (e.g., a claim line followed by the owning source link), so the deliverable format is unambiguous and copy-paste ready.
Optionally specify a default filename or frontmatter convention for the notes file when no existing convention is found, reducing ambiguity in step 3.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence — it lists what to do without explaining what a background agent or primary source is, and every line earns its place, matching the anchor 3 example. | 3 / 3 |
Actionability | The three steps are concrete and specific, but there is no example or template of the expected Markdown deliverable showing how a claim should cite its source, leaving a key detail that keeps it at anchor 2 rather than fully copy-paste ready. | 2 / 3 |
Workflow Clarity | The process is a clearly sequenced three-step list with an explicit verification requirement ('每个 claim 都要追溯到拥有该事实的 source'); since this is a non-destructive delegation task, no error-recovery feedback loop is required. | 3 / 3 |
Progressive Disclosure | The skill is under 50 lines with no need for external references and is well-organized as a brief intro plus a numbered list, so it qualifies for anchor 3 under the simple-skills allowance without bundle files. | 3 / 3 |
Total | 11 / 12 Passed |