The Agent Skills Directory

[COMMAND_EXECUTION]: The validation script scripts/quick_validate_skill.py uses subprocess.run to execute internal contract tests. This functionality is restricted to local skill verification.
[PROMPT_INJECTION]: An adversarial test fixture is included in tests/fixtures/adversarial.md. This is a defensive test case used to ensure the agent correctly prioritizes safety guidelines over user pressure to ignore red flags.
[EXTERNAL_DOWNLOADS]: The README and references link to well-known research platforms and technology organizations (such as Microsoft, Anaconda, and arXiv) for evidence mapping and design standards.
[SAFE]: No malicious patterns or unsafe data handling practices were detected. The skill includes automated checks in tests/test_skill_contract.py to prevent the accidental inclusion of credentials in project files.

production-ml-handoff