mirror of
https://github.com/meta-llama/llama-stack.git
synced 2025-12-03 09:53:45 +00:00
0e0d311dea
Add a new remote provider that integrates MLflow's Prompt Registry with Llama Stack's prompts API, enabling centralized prompt management and versioning using MLflow as the backend. Features: - Full implementation of Llama Stack Prompts protocol - Support for prompt versioning and default version management - Automatic variable extraction from Jinja2-style templates - MLflow tag-based metadata for efficient prompt filtering - Flexible authentication (config, environment variables, per-request) - Bidirectional ID mapping (pmpt_<hex> ↔ llama_prompt_<hex>) - Comprehensive error handling and validation Implementation: - Remote provider: src/llama_stack/providers/remote/prompts/mlflow/ - Inline reference provider: src/llama_stack/providers/inline/prompts/reference/ - MLflow 3.4+ required for Prompt Registry API support - Deterministic ID mapping ensures consistency across conversions Testing: - 15 comprehensive unit tests (config validation, ID mapping) - 18 end-to-end integration tests (full CRUD workflows) - GitHub Actions workflow for automated CI testing with MLflow server - Integration test fixtures with automatic server setup Documentation: - Complete provider configuration reference - Setup and usage examples with code samples - Authentication options and security best practices Signed-off-by: William Caban <william.caban@gmail.com> Co-Authored-By: Claude <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| nvidia | ||
| recordings | ||
| remote/prompts | ||
| utils | ||
| __init__.py | ||
| test_providers.py | ||