The Agent Skills Directory

[SAFE]: The skill implements strong security boundaries by explicitly instructing the agent to ignore conversational context in favor of verifiable repository evidence such as git history and source diffs.
[SAFE]: A dedicated Redaction section provides clear guidance on identifying and removing sensitive information like API keys, tokens, and personal data before saving files, and it explicitly forbids opening environment files to prevent accidental exposure.
[SAFE]: The skill limits its operational scope to three specific markdown files and suggests commit messages rather than performing autonomous git commits or network operations, maintaining human-in-the-loop control.
[SAFE]: No suspicious patterns, obfuscation, or remote code execution vectors were detected in the instructions or reference files.

project-docs