The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The web_agent_readiness.py script performs network requests using the standard library to verify the availability of agent-readiness endpoints such as llms.txt and OpenAPI specifications on target domains. This behavior is documented and necessary for the tool's core functionality.
[COMMAND_EXECUTION]: The validate_skill.py script invokes the Python interpreter via subprocess.call to execute other scripts in the package with a --help flag. This is done solely for validation purposes to ensure scripts are properly documented for agent use.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external data from file systems and websites. While this creates an attack surface, the skill is specifically an audit tool and includes detailed references on security best practices, such as references/security-recovery.md, which advises on separating untrusted data from instructions and using human-in-the-loop approvals.
[DYNAMIC_EXECUTION]: The validation tools utilize py_compile.compile to check script syntax within the skill directory. This is a standard development practice and is restricted to the local skill files during the validation workflow.

agent-use