The Agent Skills Directory

[SAFE]: No security issues detected. The skill consists entirely of natural language instructions and formatting rules for text transformation.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data in the form of user-provided commit drafts and staged changes. However, it explicitly instructs the agent not to execute Git commands, stage files, or amend changes. This lack of tool-use capability, combined with the instruction to output results only in a fenced block, effectively mitigates the risk of a prompt injection attack affecting the system or repository state.

unslop-commit