fix(models): tighten audio capability + correct openai oauth content
This commit is contained in:
@@ -89,6 +89,16 @@
|
||||
],
|
||||
"test_status": "pnpm test:run src/auth/anthropic.test.ts + pnpm typecheck passing"
|
||||
},
|
||||
"openai-oauth-and-audio-capability-tweaks": {
|
||||
"status": "completed",
|
||||
"date": "2026-02-14",
|
||||
"summary": "Fixed OpenAI OAuth (Codex) message shaping by using output_text for assistant turns, and made native-audio detection conservative via a model allowlist so Flynn does not attempt to send raw audio to models that cannot consume it.",
|
||||
"files_modified": [
|
||||
"src/models/openai.ts",
|
||||
"src/models/capabilities.ts"
|
||||
],
|
||||
"test_status": "pnpm typecheck passing (no new tests added in this change)"
|
||||
},
|
||||
"p0-p1-implementation-plan": {
|
||||
"file": "2026-02-06-p0-p1-implementation-plan.md",
|
||||
"status": "completed",
|
||||
|
||||
Reference in New Issue
Block a user