feat: add Gemini 3 thought signature support for qBraid provider #4
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
Implements proper handling of Gemini 3 thought signatures in multi-turn
function calling conversations. Gemini 3 requires thought signatures to be
passed back when sending tool results.
Changes:
- Add new qBraid provider SDK at src/provider/sdk/qbraid/ that:
- Uses a custom metadata extractor to capture _thought_signature from
tool calls in both streaming and non-streaming responses
- Stores thought signatures globally keyed by toolCallId
- Exports getThoughtSignature() for retrieval
- Update processor.ts to:
- Import getThoughtSignature from qBraid provider
- Look up thought signatures after receiving tool-call events
- Attach them to part metadata under both 'vertex' and 'google' keys
for AI SDK compatibility
- Register qBraid provider in provider.ts BUNDLED_PROVIDERS
- Update branding/qbraid/models.json to use @ai-sdk/qbraid npm identifier
so CodeQ uses our custom provider instead of generic openai-compatible
This works in conjunction with the qbraid-account proxy changes that:
1. Extract thought signatures from Gemini responses
2. Include them as _thought_signature in OpenAI-compatible tool call responses
3. Read _thought_signature from incoming requests and pass to AI SDK
|
Thanks for your contribution! This PR doesn't have a linked issue. All PRs must reference an existing issue. Please:
See CONTRIBUTING.md for details. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Summary
Implements proper handling of Gemini 3 thought signatures in multi-turn function calling conversations.
Problem
Gemini 3 models require "thought signatures" to be passed back when sending tool results in multi-turn conversations with function calling. Without these signatures, Gemini returns an error:
Solution
This PR works in conjunction with changes in qbraid-account (PR anomalyco#147) to pass thought signatures through the OpenAI-compatible API.
Changes in this PR:
New qBraid provider SDK (
packages/opencode/src/provider/sdk/qbraid/):_thought_signaturefrom tool callstoolCallIdgetThoughtSignature()for retrievalprocessor.ts updates:
vertexandgooglekeysprovider.ts:
BUNDLED_PROVIDERSmodels.json:
@ai-sdk/qbraidnpm identifierFlow:
_thought_signatureon each tool call in OpenAI-compatible responseproviderOptionsTesting
Requires qbraid-account PR anomalyco#147 to be deployed. Then test with:
codeq run --model qbraid/gemini-3-flash "can you submit a bell circuit to sv1"