Feature Request: Support custom OpenAI-compatible API endpoints (OPENAI_BASE_URL)

[rothnic]

Feature Request: Custom OpenAI-Compatible API Endpoint Support

Problem

Currently, codebase-context only supports the official OpenAI API endpoint. This limits users who want to use:

• Ollama for local LLM inference
• LiteLLM Proxy for unified model access
• Groq, OpenRouter, or other OpenAI-compatible providers
• Self-hosted models (e.g., vLLM, text-generation-inference)

Proposed Solution

Add support for the OPENAI_BASE_URL environment variable (or similar configuration) to allow users to specify a custom base URL for OpenAI-compatible API endpoints.

Expected Behavior

# Use Ollama locally
export OPENAI_BASE_URL=http://localhost:11434/v1
export OPENAI_API_KEY=ollama  # Ollama doesn't require a real key

# Or use a LiteLLM proxy
export OPENAI_BASE_URL=http://localhost:8000
export OPENAI_API_KEY=sk-...

codebase-context index /path/to/project
codebase-context ask "Explain this codebase"

Benefits

1. Privacy: Keep code analysis local with Ollama
2. Cost savings: Use local models instead of paid APIs
3. Flexibility: Use any OpenAI-compatible endpoint
4. Offline capability: Work without internet connectivity

Additional Context

I originally reported this as part of #68 (mutex lock failure on Intel Mac), but creating a separate issue to track this feature independently.

Technical Notes

• The codebase appears to use @langchain/openai which supports baseURL parameter
• This should be a relatively small change to pass through the environment variable
• Would also need to document this in the README

---

Environment

• OS: macOS Sequoia 15.2 (Intel)
• Node: v24.12.0
• codebase-context: latest

[PatrickSys]

Hey @rothnic,
Thank you for filling this issue (and #70) with such detail.

The proposed solution looks solid.
Would you be able to open a PR from your feature/openai-base-url-support branch? Happy to review and get it merged quickly.

Regards