feat(test): add orcarouter/ provider for skill-eval models - #594
Open
Marc-oss-hub wants to merge 1 commit into
Open
feat(test): add orcarouter/ provider for skill-eval models#594Marc-oss-hub wants to merge 1 commit into
Marc-oss-hub wants to merge 1 commit into
Conversation
Add OrcaRouter as a named OpenAI-compatible provider in the .test skill optimization harness, mirroring the existing Databricks AI Gateway path. - judges.py: route `orcarouter/<namespaced-model>` through https://api.orcarouter.ai/v1 with ORCAROUTER_API_KEY in both _to_litellm_model and _to_judge_model_and_params. - config.py: register context windows for orcarouter/ models; list orcarouter/ as a non-Databricks fallback in the reflection-context error message. - .test/README.md: document the orcarouter/ prefix and env vars. Co-Authored-By: Claude <noreply@anthropic.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Add
orcarouter/provider support to the skill-eval harnessThis adds OrcaRouter as a named provider for the
.test/skill-evaluation harness, mirroring how the existingdatabricks/AI Gateway routing works.OrcaRouter is an OpenAI-compatible gateway (
https://api.orcarouter.ai/v1) that serves fully-namespaced model ids (openai/gpt-4o-mini,deepseek/deepseek-v4-flash-0731, …). It also runs gateway-level, zero-trust security for AI agents on the same endpoint — screening every prompt/response and governing every tool call on a default-deny basis, with no application code changes.What changed
judges.py—_to_litellm_model()and_to_judge_model_and_params()now recognize anorcarouter/<namespaced-model>prefix (e.g.orcarouter/openai/gpt-4o-mini) and route it through OrcaRouter withORCAROUTER_API_KEYauth, exactly like the existing Databricks AI Gateway path.config.py—_register_litellm_models()registers context windows for theorcarouter/models sovalidate_reflection_context()and litellm's model registry know their limits; the "switch provider" error message now listsorcarouter/as an option..test/README.md— documents theorcarouter/prefix,ORCAROUTER_API_KEY, and exampleGEPA_REFLECTION_LM/GEPA_GEN_LMvalues.Usage
Verification
uvx ruff@0.11.0 check --select=E,F,B,PIE --ignore=E401,E402,F401,F403,B017,B904,ANN,TCH --line-length=120 --target-version=py311 .test/src/→ all checks passeduvx ruff@0.11.0 format --check --line-length=120 --target-version=py311 .test/src/→ 43 files already formatted_to_litellm_model,_to_judge_model_and_params, and model registration all behave as expected (OrcaRouter routing, key-less degradation, Databricks path unaffected).litellm.completion()call routed through the real_to_litellm_model("orcarouter/openai/gpt-4o-mini")reachedhttps://api.orcarouter.ai/v1and returnedOK.The Databricks AI Gateway path is untouched;
orcarouter/is purely additive.I'm an engineer on the OrcaRouter team.