The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes Git and the GitHub CLI (gh) through Python scripts using the subprocess module. These commands are executed in a list format, which is a secure practice to prevent shell injection. The operations include pushing branches, viewing pull request status, and creating new pull requests.
[INDIRECT_PROMPT_INJECTION]: The skill reads external data from the user's repository, including commit messages and code diffs, to generate PR descriptions. This introduces a surface where malicious repository content could influence the agent's output. \n
Ingestion points: scripts/gather_context.py reads git log and git diff output. \n
Boundary markers: None explicitly present. \n
Capability inventory: scripts/create_pr.py performs gh pr create. \n
Sanitization: None.

pr-create