Supporting OpenAI compatible endpoints in GitHubSearchTool

[agapebondservant]

Hi team,

I would like to be able to use my own custom LLM with the GitHubSearchTool and it does not appear to be working for me so far. I have a privately hosted LLM with an OpenAI-compatible endpoint deployed on vLLM (have verified that it works outside of this context). Here's the code I have:

tool = GithubSearchTool(
            github_repo=git_repo,
            gh_token=os.getenv("GH_TOKEN"),
            content_types=["code"],
            config=dict(
                llm=dict(
                    provider="openai",
                    config=dict(
                        model=my_model,
                        api_key=my_api_key,
                        base_url=my_base_url,
                        temperature=0.1,
                    ),
                ),
            )
        ))

When I attempt to invoke the crew, I get

Error invoking Github Query Agent: Error code: 401 - {'error': {'message': 'Incorrect API key provided: v1*************************************************************2483. You can find your API key at https://platform.openai.com/account/api-keys.', 'type': 'invalid_request_error', 'param': None, 'code': 'invalid_api_key'}} in upsert.

Seems it doesn't recognize my base URL which is not OpenAI (it's just OpenAI compatible). What am I missing here? Any help would be greatly appreciated - thanks!