Mindy MCP exposes 2 tools via HTTP Streamable transport. Both tools have descriptions and basic input schemas. get_balance has an empty input schema ({}), which is appropriate for a parameterless tool. search_sam_opportunities has a properly typed schema with 2 string parameters. However, parameter descriptions are minimal (under 50 chars each), and output schemas are not documented in the source. No error handling guidance, no tool annotations, and no indication of pagination support for search results. The server follows basic naming conventions (verb_noun) but lacks depth in parameter documentation and output structure guidance.

• Document the output schema for search_sam_opportunities. Specify fields returned (e.g., opportunity_id, title, agency, deadline, set_aside, url) with types and descriptions. Include a note on result limits (e.g., 'Returns up to 50 opportunities per call').
• Expand parameter descriptions to 50-100 chars. For 'set_aside', list valid values: 'Set-aside category filter. Valid values: SBA, HUBZone, VOSB, WOSB, or leave empty for all.' For 'keyword', clarify format: 'Search keyword (1-100 chars, case-insensitive). E.g., "IT staffing", "cloud services".'
• Add pagination support. Include limit (default 20, max 100) and offset or cursor parameters in search_sam_opportunities. Return total_count and next_cursor in the response so agents can iterate through large result sets.
• Add error handling guidance in descriptions. E.g., 'If no opportunities match, try broadening the keyword or removing the set_aside filter. Call again with a simpler keyword.'
• Declare tool annotations in the schema. Mark both tools with readOnlyHint: true to signal to LLMs that they are safe to call without user confirmation.