recommender

Analyzes session changes to detect missing verification skills. Dynamically explores existing skills, creates new skills or updates existing ones, and manages CLAUDE.md.

Checks if code could be simplified — redundant booleans, manual loops, verbose patterns, if-else chains replaceable by dict lookups. Use after adding or modifying Python code.

Sequentially executes all verify skills in the project to generate an integrated verification report. Use after feature implementation, before PRs, or during code review.

Verifies new models are registered in ModelName enum, IMPLEMENTED_MODELS, MODEL_PATH, and have proper train.py routing. Use after adding a new model.

Verifies PEP 8 naming, type hints, ruff compliance, import ordering, and project-specific conventions. Use after adding or modifying Python code.

Verifies that new models and pipeline components have corresponding unit tests and integration tests. Use after adding new models or modifying training pipeline.