The Agent Skills Directory

[SAFE]: The skill performs a legitimate utility task using standard local tools. No malicious behaviors such as unauthorized data exfiltration, remote code execution, or obfuscation were detected.
[INDIRECT_PROMPT_INJECTION]: The skill includes a dedicated Security Model section that explicitly instructs the agent to treat all scanned content as data and never follow instructions found within the project files. This mitigation, combined with the narrow character-replacement scope, addresses the risk associated with processing untrusted text.
[PERSISTENCE]: The skill provides a git commit-msg hook to maintain style consistency in repository history. This use of persistence is a standard development practice, is clearly documented for its specific purpose, and requires manual installation by the user.

dashfix