Single tool 'introbot_scan' has a reasonable description (127 chars, within 10-1024 range) but critical schema deficiencies. Input schema is minimal: only 'target' (string) with basic description, no constraints (enum, pattern, min/max length). No output schema documented, LLMs cannot predict return structure. Parameter description lacks actionable detail (format, examples of valid targets, expected behavior). Tool name 'introbot_scan' uses verb_noun pattern correctly but description does not clarify when to use it vs. other discovery tools or what 'warm-intro paths' means to an LLM unfamiliar with the domain. No error handling guidance, no pagination info, no security/permission notes.

• Document the complete output schema in the tool description or as a JSON Schema response type. Example: 'Returns {"paths": [{"person": string, "connection_chain": string[], "strength": number}], "total_found": number}'.
• Add constraints to the 'target' parameter: describe valid formats (e.g., 'Email address or full name of the target person'), provide min/max length, and list any disallowed characters.
• Expand the tool description to explain domain context: 'Finds introduction paths through your team's professional network. Requires a contacts manifest (CSV/JSON) loaded beforehand. Returns ranked paths with connection strength scores.'
• Add error handling guidance: 'If no paths found, try a broader target name or verify the contacts manifest is loaded. If network error, retry after 30 seconds.'
• Migrate from STDIO to HTTP or HTTP+SSE transport to enable remote access and integration with hosted MCP clients.