The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted text from external files or user input, which is then used to generate file modifications via patch or write_file. This represents a standard indirect prompt injection surface; however, the skill explicitly instructs the agent to show diffs and changed sections to the user for review before applying edits, which serves as a human-in-the-loop safety boundary.
[SAFE]: The skill performs text analysis and editing using standard file system tools. It does not initiate network connections, execute remote code, or request elevated privileges. The referenced external repository is used solely for attribution and does not involve automated downloads or script execution.

humanizer