The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to install standard Python libraries (mlflow[extras], psycopg2-binary, boto3) required for model registry management and database communication.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data in the form of model metadata and binary artifacts, which are used to update registry tags and metrics.
Ingestion points: register_model.py ingests a metadata dictionary and reads the artifact at model_path; promote_model.py takes command-line parameters for version and approver info.
Boundary markers: The skill implements Pydantic-based schema validation and provides examples for OPA policy enforcement.
Capability inventory: The scripts interact with the filesystem (reading artifacts), perform cryptographic hashing (SHA-256), and communicate with a remote MLflow tracking server via HTTP.
Sanitization: Inbound metadata is validated against the ModelMetadata schema to ensure data integrity before registration.

model-registry-governance