urlscan.io URL scanner — search/result keyless, submit needs key
This server has critical gaps in parameter descriptions and output schema documentation. While tool names are action-oriented (search, result, submit, domain, ip), most parameters lack descriptions entirely. The 'submit' tool has some parameter descriptions, but search, result, domain, and ip have zero parameter documentation. No output schemas are documented anywhere. Error handling exists but is minimal. The server lacks pagination guidance, field naming consistency, and LLM-optimized descriptions. Most tools fall well below production baseline (194 chars avg description, 100% param coverage in A+ tools).
Search by domain.
Search by IP.
Full result by scan uuid (keyless).
Search past scans (keyless).
Submit a new URL for scanning (requires key).
Missing parameter descriptions on 4 of 5 tools. 'search' lacks descriptions for 'size' and 'search_after'; 'result', 'domain', 'ip' have zero param docs. LLMs cannot infer parameter meaning from names alone.
No output schemas documented. Tools return raw API responses (JSON) but LLMs have no guidance on expected fields, types, or structure. Prevents downstream tool chaining and forces LLM to guess field names.
Pagination not documented. 'search' accepts 'size' and 'search_after' but no description explains pagination behavior, limits, or when to use search_after. LLMs cannot plan multi-page queries.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 46 | 2026-07-28+ | v2 |
Tool descriptions are too brief (10-65 chars). Baseline is 194 chars. 'Search past scans (keyless)' and 'Full result by scan uuid (keyless)' lack context on WHEN to use each tool vs alternatives, what data is returned, and prerequisites.
API key handling via _apiKey parameter is non-standard. While the code checks for PLATFORM_URLSCAN_KEY env var, exposing _apiKey as a parameter violates secret-injection pattern. Credentials should never be tool parameters.