What problem does it solve? It standardizes how quantitative model pull requests are reviewed, validated, and handed off, ensuring every shared model entry carries real metadata, passes contract validation, and is written with explicit version-controlled approval. ## Core Features & Use Cases - PR Metadata Extraction: Reads authorized GitHub PRs and commits to collect training data timestamps, model type, hyperparameters, feature dependencies, and risk metadata without fabricating defaults. - ModelSpec Contract Validation: Runs candidate metadata through the published generate_model_spec tool for real schemas.model.ModelSpec validation rather than trusting model-generated content. - Versioned Shared Handoff: Reads the current shared.model_entries blackboard version and proposes writes with expected_version so merges require explicit user approval. - Use Case: A quant researcher finishes a factor model PR and needs it registered for the risk team; this Skill validates the ModelSpec, freezes the task plan, and completes the shared handoff only after the user approves the exact merge. ## Quick Start Ask the agent to process an authorized model PR, validate its ModelSpec metadata, and submit it to the shared model entries blackboard for approval.
