Repository navigation
feat(python): connect self-hosted chat models as policies - #71
Merged
Merged
Conversation
nosam1998
marked this pull request as ready for review
October 1, 2026 02:58
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Connect a caller-owned model through an OpenAI-compatible chat endpoint. ChatPolicy sends only the acting seat's observation, game rules and legal action strings, validates structured replies against the legal mask, and maps them to native action indices. Credentials come from a named environment variable; failures have stable redacted codes and Gym rolls back failed exchanges.
Validation: all eight CI workflows pass, including Linux/Windows/macOS HTTP protocol, malformed-response, authentication, timeout, redirect, size-limit and rollback tests. Tests use local HTTP fixtures; no paid model calls are made.