The Agent Skills Directory

[PROMPT_INJECTION]: The skill contains an explicit instruction to 'Run autonomously without approval'. This is a safety bypass pattern that attempts to remove human-in-the-loop oversight for commands that modify the file system, increasing the risk of unauthorized or harmful persistent changes.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface. Ingestion points: Untrusted data is ingested via the {{INPUT}} placeholder and by analyzing staged/unstaged changes in the repository (SKILL.md). Boundary markers: There are no delimiters or instructions to treat external data as non-executable content. Capability inventory: The agent is authorized to use git add and git commit to modify the repository (SKILL.md). Sanitization: No sanitization or validation is applied to the analyzed file content or input data.
[COMMAND_EXECUTION]: The skill utilizes shell commands (git add, git commit) to modify the local environment. When combined with the autonomous execution instruction, this creates a significant risk that a malicious payload in a file change or user input could trick the agent into committing harmful code or sensitive data without user verification.

cmd-git-commit