The Agent Skills Directory

[SAFE]: The skill is a documentation classification and writing assistant. Analysis of all provided scripts and instructions confirms that the skill performs legitimate documentation tasks without any signs of data exfiltration, prompt injection, or obfuscation.
[COMMAND_EXECUTION]: The scripts scripts/check_local.py and tests/test_audit_docs.py utilize subprocess.run() to execute Python unit tests. This is a standard developer practice for validation and testing within the local environment and does not involve executing untrusted user input.
[EXTERNAL_DOWNLOADS]: The installation guides for Claude Code and OpenCode reference standard git clone commands to download the skill repository from its official GitHub home. These downloads are transparently documented and follow established patterns for agent skill installation.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a natural attack surface for indirect prompt injection as it is designed to ingest and analyze documentation content provided by the user. While the skill lacks explicit boundary markers for untrusted content, its capabilities are limited to local documentation analysis and validation, presenting no significant security risk.

diataxis-docs