The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes standard, read-only Git commands (git status, git diff --cached, and git log) to analyze staged changes and commit history. These operations are essential to its function of generating context-aware commit messages.
[PROMPT_INJECTION]: While the skill ingests untrusted content via git diff (which could contain malicious instructions in code comments), this represents a standard risk for development tools. The skill's instructions focus strictly on generating a commit message, which limits the impact of such an attack surface.
Ingestion points: Output from git diff --cached and git log described in SKILL.md.
Boundary markers: None identified.
Capability inventory: Limited to read-only git operations.
Sanitization: None identified.
[SAFE]: No network requests to external domains, hardcoded credentials, or dynamic code execution patterns were found.

gcm