How to design MCP tools for agents
MCP tools are the operations an MCP server exposes for a model to call. Each has a name, a description and a JSON Schema for its inputs, and the model decides when to call one, so the model reads the tool list before every answer. Design them around the questions your users ask rather than the endpoints a vendor ships: fewer, larger tools, and one consolidation layer in front when two or more…
Designing MCP tools for agents involves creating a set of operations that an MCP server exposes for a model to call. These tools should be designed based on the questions users ask rather than the endpoints provided by vendors. The recommended approach is to have fewer, larger tools and a consolidation layer that sits in front of multiple sources, deciding which source is authoritative for each question.
This consolidation layer should return structured JSON, authenticate every request as the user, and log every call. Direct vendor wires are reasonable for a single read-only source with a single owner, but become unreasonable once a second source for the same question arrives. The consolidation layer should handle authentication, rate limits, and logging, and decide the next change based on the call log.
Brief written by urgent.news from Dev.to's own syndicated text. Machine-written — may contain errors; check the original before relying on it.
