Skip to content

Add opt-in observation of physical model requests - #585

Open
flamerged wants to merge 1 commit into
LibreChat-AI:mainfrom
flamerged:feature-model-attempt-observation
Open

flamerged wants to merge 1 commit into
LibreChat-AI:mainfrom
flamerged:feature-model-attempt-observation

Conversation

@flamerged

Copy link
Copy Markdown

Adds an opt-in observer at the physical fetch boundary for OpenAI-compatible transports. It reports bounded routing, usage and terminal metadata without reading request bodies or changing response bytes, backpressure, cancellation or retry policy. Ambiguous termination remains unknown, and stream charges retain their source.

Exports the helper and passes OpenAI ClientOptions through OpenRouter. It makes no additional model or accounting requests.

Validation: focused transport/privacy/native-client suite passed 10 tests; TypeScript/package build and scoped lint passed; built CommonJS/native HTTP smoke preserved output across three actual loopback dispatches and verified retry counts, usage projection and private-marker exclusion. No live-provider or downstream task-success claim is made.

This branch has not been deployed

No deployments
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.

1 participant