This server has severe definition quality issues. The single tool `execute_skill_action` has a reasonable schema with typed parameters and a basic description, but the tool itself is fundamentally broken as an MCP implementation. The schema shows three parameters (target_niche_topic, min_followers, target_geo_distribution) with types and defaults, but critical gaps exist: (1) the description is minimal (59 chars) and lacks context about when/why to use the tool or what the business outcome is; (2) there is NO documented output schema, the mcp_server.py only prints a static JSON response with 'supported_tools' but never actually implements tool execution or returns results; (3) parameter descriptions are present but generic ('Target niche topic for creator matching') without guidance on valid values, constraints, or dependencies; (4) the actual implementation in client.py is a mock that returns hardcoded results and never validates inputs; (5) no error handling exists, the server will crash on invalid input with no recovery guidance; (6) the tool combines multiple concerns (niche matching + follower filtering + geo distribution) which violates single-responsibility principle. The code shows the tool is non-functional as an MCP server, mcp_server.py does not implement the MCP protocol's tool execution flow, it only announces capability but cannot actually execute tools.

• Rewrite mcp_server.py to implement the MCP protocol's request/response loop. Currently it only prints a static message on startup. It must: (1) listen for tool calls via stdin, (2) parse the request JSON including tool name and parameters, (3) invoke the tool (or client), (4) serialize the response (including documented output schema), (5) write it to stdout. See https://spec.modelcontextprotocol.io/latest/basic/protocol-overview/ for the full request/response format.
• Expand tool description to 80 - 150 characters explaining: WHAT (find creators in a niche matching an audience profile), WHEN (use this before launching a partnership or campaign), WHAT YOU NEED (a niche topic, min follower threshold, target geo), WHAT YOU GET (a list of matching creators with engagement metrics). Example: 'Find creators in a target niche with specified follower counts and geographic audience distribution. Use this to identify partnership candidates or lookalike audiences for influencer campaigns.'
• Document the output schema explicitly in the tool definition. Add a 'output' field to the tool registration or include it in the description: 'Returns: { matching_session_id (string), creators_discovered_count (integer), top_tier_overlap_similarity_pct (number 0-100), avg_engagement_rate_pct (number 0-100), creator_roster_manifest_url (string, URL to full roster) }'
• Convert 'target_niche_topic' from free-form string to an enum with all valid niche options. If niches are dynamic, document the discovery mechanism (e.g., 'Call list_available_niches() first to see valid values').