The Agent Skills Directory

[SAFE]: The skill implementation follows secure development practices by using the Python standard library exclusively, avoiding third-party dependency risks and supply chain vulnerabilities.
[SAFE]: Implements a hardened HTTP client with a custom redirect handler in scripts/arax_client.py that enforces same-origin and protocol-consistency (HTTPS), effectively mitigating potential Man-in-the-Middle and redirection-based attacks.
[SAFE]: Rigorous input sanitization and identifier validation using regular expressions (CURIEs, Biolink Model terms) protect against malformed data and potential injection into the ARAXi operation language.
[SAFE]: Secure artifact persistence patterns, including atomic write operations on temporary files followed by an atomic move and restrictive file permissions (0600), protect local data integrity and confidentiality.
[SAFE]: Explicit safety boundaries and operational constraints are defined in SKILL.md and references/query-contract.md to prevent the submission of sensitive patient or proprietary research data to public endpoints.

ncats-arax