weijiafu14 8bfdbda251 fix: inject thoughtSignature into functionCall parts for OpenAI→Gemini conversion ...

When requests arrive via OpenAI-compatible endpoints and are converted to
Gemini format, functionCall parts were missing thoughtSignature. Gemini 3+
rejects such requests with 400 when thinking mode is enabled.

Changes:
- Extract thoughtSignature from tool_calls provider_specific_fields
- Inject into corresponding functionCall parts
- Fall back to dummy sentinel when thinking is enabled but no signature available
- Split signed/unsigned parts into separate model messages (Gemini requirement)

All existing tests pass (59/59).

2026-03-17 16:08:08 +08:00

..

desktop

feat: add masked token placeholders and desktop icons

2026-03-16 21:36:23 +08:00

server

fix: inject thoughtSignature into functionCall parts for OpenAI→Gemini conversion

2026-03-17 16:08:08 +08:00

web

feat: add multi-provider OAuth support for Claude and Gemini CLI

2026-03-17 13:36:44 +08:00

encoding.mojibake.test.ts

feat: update error messages in NewApiAdapter tests to use correct Chinese translations

2026-03-06 17:10:01 +08:00