Introduces AIMLAPI as a supported AI provider, including model list, config validation, and engine implementation. Updates README and engine selection logic to integrate AIMLAPI for chat completions.
Refactor AimlApiEngine response handling
Removed dependency on removeContentTags and simplified message content extraction. Minor header formatting fix for HTTP-Referer. This streamlines the response handling and reduces unnecessary processing.
The changes:
1. Expanded the skip condition to include additional AI providers (DEEPSEEK, GROQ, MISTRAL, MLX, OPENROUTER) beyond just TEST
2. Maintained existing TEST provider skip behavior
3. Added explicit comment explaining the skip logic
The why:
Prevents migration execution for unsupported AI providers to avoid potential runtime errors or data inconsistencies, ensuring migrations only run for properly handled configurations.
- Add prettier job to GitHub Actions workflow to enforce code formatting
- Add format:check script to package.json for checking formatting
- Include failure message when prettier check fails in CI
Add OpenRouterEngine class and integrate it into the configuration
system. OpenRouter provides access to 300+ AI models through a
unified API, expanding model availability for commit message
generation beyond existing providers.
When user wants one line commits the system prompt and the user/assistant one-shot example were contradicting each other, confusing the LLM. This fix modifies the assistant output so that prompt and one-shot are consistent.
This commit adds a new 'describe' mode to the config command, allowing users
to get detailed information about configuration parameters. It includes:
1. New CONFIG_MODES.describe enum value
2. Functions to generate and print help messages for config parameters
3. Updated configCommand to handle the new 'describe' mode
4. README updates to document the new 'describe' functionality
- Added OCO_GITPUSH='true' environment variable to all e2e test commands
- Fixed ESM module import issues in commitlint tests by using process.cwd()
- Replaced import.meta.url usage with relative paths from project root
- All e2e tests now properly test the git push prompt functionality
- Fixed Jest configuration error by removing duplicate modulePathIgnorePatterns property
- Consolidated the ignore patterns into a single declaration
- This resolves the TypeScript compilation error preventing tests from running
- export parseCustomHeaders from src/utils/engine.ts
- use parseCustomHeaders in OpenAiEngine for config.customHeaders
- remove try/catch and inline JSON.parse logic
- update config test to expect headers as object and drop JSON.parse
Centralize header parsing for reuse and simplify engine code
Update tests to match new header format for clarity
Use OpenAI.ClientOptions for stronger typing and clarity
Extract custom headers parsing into parseCustomHeaders util
Simplify getEngine by delegating header parsing to helper
Improve maintainability and reduce code duplication
Add OCO_API_CUSTOM_HEADERS variable to README, config enum,
and env parsing to allow JSON string of custom headers.
Validate that custom headers are valid JSON in config validator.
Extend AiEngineConfig with customHeaders and pass headers to
OllamaEngine and OpenAiEngine clients when creating requests.
Parse custom headers in utils/engine and warn on invalid format.
Add unit tests to ensure OCO_API_CUSTOM_HEADERS is handled
correctly and merged from env over global config.
This enables users to send additional headers such as
Authorization or tracing headers with LLM API calls.
