Skip to content

Conversation

@Tarquinen
Copy link
Collaborator

@Tarquinen Tarquinen commented Jan 16, 2026

Summary

  • Uses tool parts with skip_thought_signature_validator bypass for Gemini instead of text injection
  • Provides consistent tool part injection across all providers
  • Simplifies code by removing Gemini-specific text part branch

Tarquinen and others added 5 commits January 14, 2026 01:43
v1.2.1 - Fix Claude reasoning block injection
Instead of using text injection for Gemini models, use tool parts with
the 'skip_thought_signature_validator' bypass signature. This flows
through: ToolPart.metadata -> callProviderMetadata -> providerOptions
-> @ai-sdk/google reads providerOptions.google.thoughtSignature.

This provides consistent tool part injection across all providers while
avoiding Gemini's thought signature validation errors.
@Tarquinen Tarquinen merged commit fb662af into dev Jan 16, 2026
1 check passed
@Tarquinen Tarquinen deleted the feat/gemini-thoughtsignature-bypass branch January 16, 2026 19:33
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants