The Agent Skills Directory

[COMMAND_EXECUTION]: The script fix_gdocs_footnotes.py executes a peer script create_crossrefs.py using subprocess.run. This is a controlled internal operation used to chain steps in the document restoration workflow and does not allow for arbitrary command injection.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to the Google Gemini API for processing footnotes and the Perma.cc API for archiving legal URLs. Both are established, well-known services directly related to the skill's stated purpose of legal document auditing.
[DATA_EXFILTRATION]: While the skill transmits document fragments (footnotes) and URLs to external APIs, these operations are integral to the auditing and archiving features. No unauthorized data exfiltration to suspicious or non-whitelisted domains was detected.
[PROMPT_INJECTION]: The skill uses emphatic instructional language (e.g., "Iron Law", "EXTREMELY-IMPORTANT", "Rationalization Table") and "STOP" triggers to ensure the AI agent adheres to a strict linear workflow and does not skip validation steps. These are legitimate procedural safeguards for complex tasks and do not represent a security risk.

bluebook-audit