The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands including git diff, git commit, and gh pr edit to inspect changes and apply remediations. It accepts a argument that is interpolated into a git command, which could pose a risk if inputs are not sanitized.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from git diff output (specifically comments, docstrings, and changelog entries) as described in the SKILL.md scope section. There are no explicit boundary markers or delimiters defined to isolate this data from the agent's instructions. The agent possesses capabilities to edit files and modify PR metadata based on its findings. No specific sanitization or filtering of the ingested content is described.

prose-review