Tool Schema Design

Design LLM-callable tool signatures and JSON Schemas so the model picks the right tool and supplies valid arguments. Use when writing or reviewing function/tool definitions for OpenAI, Anthropic, Google, Bedrock, MCP servers, or any agent framework (LangChain, LangGraph, Semantic Kernel, Claude Code, OpenCode, Pi). Covers tool descriptions (what + when + when-NOT-to-use), parameter descriptions with format/range/example, JSON Schema constraints (enum, minimum, format), required-minimalism, naming consistency across a toolkit, structured outputs with strict schemas, tool_choice, parallel calls, and error responses that steer the model to recover. For portable Agent Skills use skill-creator; for MCP server architecture use mcp-creator.

AeonDave Updated 37 repo stars

File contents

AeonDave/malskill commit 1e54873b32

Frequently asked questions

npx get-skill add aeondave/tool-schema-design