myco:mcp-tool-development-lifecycle

Automating the MCP tool development lifecycle involves enforcing schema definitions, deterministic handler patterns, and documentation bundling to prevent drift. This guarantees agents always have aligned tool invocations by validating the schema-handler-documentation triad during the build process.

To prevent documentation drift, bundle agent-visible docs at build time and run anti-drift tests validating schema, handler, and documentation alignment. This catches synchronization errors between tool definitions and server registration before they reach production.

Register an MCP tool in server.ts using a switch-based pattern by mapping tool names to handler implementations. Combine this with TOOL_DEFINITIONS schemas and DaemonClient usage to ensure deterministic I/O across local and collective surfaces.

Choose cloud versus local placement for MCP tools based on data exposure requirements and collective enablement needs. Local placement prevents unintended data exposure, while cloud placement supports broader access when documentation and handler alignment is confirmed.

Yes, deterministic MCP tool handler patterns require DaemonClient to enforce strict parameter consumption. This guarantees deterministic I/O and maintains the schema-handler alignment needed for reliable agent tool invocations across surfaces.

MCP tool schema drift from handler implementations happens when build-time documentation bundling and anti-drift tests are skipped. Applying lifecycle procedures for schema definitions and server.ts registration prevents this misalignment over time.