MCP Tool Schema Validator

MCP is an open protocol by Anthropic that standardizes how AI assistants connect to external tools, data sources, and services. It defines a JSON-based interface where servers expose 'tools' (actions), 'resources' (data), and 'prompts' (reusable templates) that LLM clients can discover and invoke.

A valid MCP tool must have: a 'name' (string, required), a 'description' (string, required), and an 'inputSchema' (object with type: 'object' and a 'properties' object). The inputSchema follows JSON Schema draft-07. Optionally include a 'required' array listing mandatory properties.

MCP tool property types must be valid JSON Schema primitive types: 'string', 'number', 'integer', 'boolean', 'array', or 'object'. You can also use 'null'. Complex types use 'type' arrays (e.g., ['string', 'null'] for nullable fields) or $ref for reusable definitions.

Tool names should be lowercase with underscores (snake_case), descriptive, and action-oriented. Good examples: 'read_file', 'create_calendar_event', 'search_documents'. Avoid spaces, special characters, or overly generic names like 'do_thing'. The name is used by the LLM to decide which tool to call.

The 'required' array lives inside inputSchema (not at the top level) and lists property names that the caller MUST provide. Properties not in 'required' are optional and the tool handler should provide sensible defaults. Always validate required fields server-side even if marked in the schema.