# Agent Tool Schema Validation
Status: public
Confidence: medium (0.865) (verified)
Last verified: 2026-06-02
Generation: ai_structured


## TL;DR

Agent tool schema validation is the practice of defining, constraining, and checking the structured inputs and outputs that flow between an LLM agent and external tools.

## Core Explanation

Tool schemas are a reliability boundary. They tell the model what arguments a tool accepts, help the runtime reject malformed calls, and give downstream code a predictable output contract. In production agent systems, schemas should be versioned, validated before execution, validated again after tool return, and paired with separate authorization and side-effect policies.

## Source-Mapped Facts

- The Model Context Protocol tools specification says a tool definition includes an inputSchema JSON Schema for expected parameters and an optional outputSchema JSON Schema for expected output structure. ([source](https://modelcontextprotocol.io/specification/2025-06-18/server/tools))
- JSON Schema Validation Draft 2020-12 specifies a vocabulary for JSON instance validation and assertions about what a valid document must look like. ([source](https://json-schema.org/draft/2020-12/json-schema-validation))
- OpenAI function calling documentation says setting strict to true makes function calls reliably adhere to the function schema instead of being best effort. ([source](https://platform.openai.com/docs/guides/function-calling?api-mode=chat))

## Further Reading

- [MCP tools specification](https://modelcontextprotocol.io/specification/2025-06-18/server/tools)
- [JSON Schema Validation Draft 2020-12](https://json-schema.org/draft/2020-12/json-schema-validation)
- [OpenAI function calling](https://platform.openai.com/docs/guides/function-calling?api-mode=chat)