The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local version control commands including git status, git diff, git add, and git commit to automate development workflows.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface by reading repository diffs that could contain malicious instructions.
Ingestion points: The skill reads untrusted data via git diff --cached in the analysis phase.
Boundary markers: No specific delimiters or safety warnings are used to isolate the diff content from the agent's logic.
Capability inventory: The skill has permissions to stage files and execute commits with arbitrary messages.
Sanitization: No sanitization is applied to the diff data before it influences the generated commit message.
[SAFE]: No evidence of malicious behavior, data exfiltration, or obfuscation was found. The core functionality is restricted to local git operations, and the mandatory user confirmation step for the proposed commit message provides an effective mitigation against unintended actions.

commit-smart