The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to perform standard Git operations.
It uses commands such as git diff, git status, git add, and git commit to manage repository state.
A "Git Safety Protocol" is implemented to mitigate risk, explicitly prohibiting the agent from modifying Git configuration, skipping hooks, or performing destructive actions like --force or hard reset without explicit user request.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection attack surface due to its analysis of untrusted repository data.
Ingestion points: The agent reads potentially untrusted code content through git diff and git status --porcelain as described in SKILL.md.
Boundary markers: The provided instructions do not specify the use of delimiters or isolation techniques to separate code diffs from the agent's internal instructions.
Capability inventory: The agent can perform file system staging and commit operations via the Bash tool.
Sanitization: There are no documented steps for sanitizing or escaping the content of the diffs before they are processed by the LLM for message generation.

git-commit