Content
92%Weight 40%Scale 1-3Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A tight, actionable single-file workflow with executable code, clear sequencing, and recovery guidance; its only weakness is progressive disclosure, since reference material is inlined instead of split into bundled files.
Suggestions
Move the Search Types table and Error Handling table into a bundled references file (e.g. references/api-reference.md) and link to it, keeping SKILL.md as a concise workflow overview.
Extract the full parameter enumerations (all category values, livecrawl/maxAgeHours modes, numResults limits) into references/ so the core instructions stay compact.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean — overview, a compact search-types table, five short step sections with executable code, and an error table — with prose limited to Exa-specific behavior Claude would not already know. It assumes competence and avoids explaining generic concepts. | 3 / 3 |
Actionability | Each step ships complete, executable TypeScript with concrete parameters (numResults, type, date filters, categories), and the error table maps HTTP codes to specific fixes — copy-paste ready. | 3 / 3 |
Workflow Clarity | Steps 1–5 are clearly sequenced from basic search to freshness control, and the Error Handling table supplies cause/solution pairs that function as recovery feedback loops. | 3 / 3 |
Progressive Disclosure | Sections are well-organized and external docs are linked, but at ~190 lines with no bundle files the Search Types reference table and Error Handling codes live inline rather than in a split reference file, so the overview is not as minimal as it could be. | 2 / 3 |
Total | 11 / 12 Passed |