OSINT MCP Server - Open Source Intelligence gathering tools via Model Context Protocol
Scoring was not performed
Output schemas not documented. Most tools do not declare their return type structure, only input schemas are visible. LLMs cannot plan downstream tool calls or extract relevant fields without knowing what response fields to expect.
Parameter descriptions lack actionable constraints. Descriptions like 'Domain name to find subdomains for' or 'Email address to validate' omit format requirements, validation rules (e.g., regex, length), range constraints for numerics, or examples of valid input.
No dependency hints or disambiguation between similar tools. Tools like 'lookup_btc' and 'lookup_eth' lack descriptions explaining when to use BTC vs ETH. Tools like 'discord_lookup', 'steam_lookup', 'xbox_lookup' do not clarify which platform applies or how to resolve a username to a platform-specific ID.
Pagination and result limiting not consistently visible. Tools like 'github_user_repos', 'fandom_user_contributions', 'exa_search' accept a 'limit' parameter, but schema/description does not clarify whether pagination (offset/cursor) is supported or what the maximum results cap is.
Error messages not actionable. BaseApiClient throws McpError with InternalError code and message like 'API request failed: 404 Not Found' or 'Rate limit exceeded'. These do not tell the agent what to do next (retry? lookup? ask user?).
Tool descriptions missing rationale for use. Tool descriptions state WHAT the tool does but not WHEN to use it or why one tool is chosen over a similar one. E.g., 'search_emails_by_domain' vs 'email_permutations', when should an agent pick each?
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 13 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 37 | - | v1 |