🤖 fix: sanitize MCP tool schemas for OpenAI Responses API compatibility #1220
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
OpenAI's Responses API has stricter JSON Schema validation than other providers like Anthropic. Certain schema properties that are valid JSON Schema but not supported by OpenAI cause 400 errors when MCP tools are used. This change adds a schema sanitizer that strips unsupported properties from MCP tool schemas when using OpenAI models: - String validation: minLength, maxLength, pattern, format - Number validation: minimum, maximum, exclusiveMinimum, exclusiveMaximum - Array validation: minItems, maxItems, uniqueItems - Object validation: minProperties, maxProperties - General: default, examples, deprecated, readOnly, writeOnly The sanitization is applied only to MCP tools and only for OpenAI models, preserving the full schema for other providers.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
MCP tools from @ai-sdk/mcp use inputSchema with a jsonSchema getter, not parameters like regular AI SDK tools. The sanitizer now handles both cases to properly strip unsupported properties for OpenAI.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Problem
OpenAI chats were unable to use MCP tools because OpenAI's Responses API has stricter JSON Schema validation than other providers. MCP servers may define tool schemas with properties like
minLength,maximum,default, etc. which are valid JSON Schema but cause 400 errors with OpenAI:Solution
Added a schema sanitizer that strips unsupported JSON Schema properties from MCP tool schemas when using OpenAI models. The sanitization is only applied to:
Properties stripped:
The sanitization preserves the core schema structure (type, properties, required, etc.) and handles nested schemas including items, oneOf, anyOf, allOf, and definitions.
Testing
Generated with
mux• Model:anthropic:claude-opus-4-5• Thinking:high