The Agent Skills Directory

[SAFE]: The skill provides documentation and code examples for the LaminDB framework. All external references are directed toward official documentation, well-known GitHub repositories, and trusted cloud infrastructure providers (AWS, Google Cloud, Cloudflare).
[COMMAND_EXECUTION]: The skill includes legitimate command-line instructions for environment setup, package installation via standard managers (pip, uv), and framework management through the LaminDB CLI.
[CREDENTIALS_UNSAFE]: No hardcoded credentials or secrets were found. The documentation follows security best practices by instructing users to utilize environment variables (e.g., AWS_ACCESS_KEY_ID) and secure CLI authentication workflows for managing access to cloud storage and databases.
[DATA_EXFILTRATION]: No unauthorized data exfiltration patterns were detected. The skill describes intended data movement between local environments and user-controlled cloud storage as part of its core functionality.
[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it is designed to ingest and process external biological datasets.
Ingestion points: Processes various file formats via pandas.read_csv, anndata.read_h5ad, and the LaminDB Artifact.load method across several reference files.
Boundary markers: No explicit boundary markers or isolation instructions are present in the snippets for data loading.
Capability inventory: Capabilities include file system writes, cloud storage synchronization, and database record creation.
Sanitization: The skill mitigates risks by implementing rigorous schema validation and biological ontology standardization (via Bionty) to ensure data integrity and structural compliance.

lamindb