Content
57%Reviews 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 contains genuinely SearXNG-specific guidance, but it pads in generic research advice Claude already knows and, critically, never explains how to actually call a SearXNG instance. Tightening the generic sections and adding the invocation mechanism would raise it to the top band.
Suggestions
Add the concrete SearXNG invocation mechanism — the instance URL pattern, HTTP endpoint, and the actual query parameters (q, categories, pageno) — so guidance is copy-paste executable.
Trim or remove generic advice Claude already knows (e.g. 'Lead with the direct answer', 'Organize findings by relevance') to improve token efficiency.
Add an explicit numbered search workflow with a verification checkpoint, e.g. 1) pick category, 2) form query, 3) run search, 4) verify sources resolve before citing.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The SearXNG-specific material (category table, !engine syntax, site: and quote operators) earns its place, but the 'Synthesizing Results' and 'Pitfalls' sections repeat general research practices Claude already knows ('Lead with the direct answer', 'Organize findings by relevance'), so it could be tightened. | 2 / 3 |
Actionability | It gives concrete query techniques (e.g. '!wikipedia rust programming', 'site:example.com', '"rust borrow checker"') but never specifies how to actually invoke SearXNG — no endpoint URL, request format, or parameter names — leaving a key execution detail missing. | 2 / 3 |
Workflow Clarity | The section order loosely sequences a search workflow (category → query → execute → synthesize), but there are no explicit numbered steps or validation checkpoints, so the sequence is implicit rather than clearly sequenced. | 2 / 3 |
Progressive Disclosure | It is a single-purpose, self-contained skill with well-organized headers, a compact reference table, and no nested or external references — content is appropriately inline and easy to navigate. | 3 / 3 |
Total | 9 / 12 Passed |