The Agent Skills Directory

[COMMAND_EXECUTION]: The skill automates repository workflows by executing multiple shell commands, including git status, git fetch, gh pr list, and gh pr create. These tools are used to manage pull requests and compare local and remote repository states.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it ingests and acts upon data from the external environment (e.g., existing pull request details and file system state).
Ingestion points: Data enters the agent context through the outputs of gh pr list (PR titles/bodies) and git status (file names/paths).
Boundary markers: The skill utilizes a structured markdown template (pr-body-template.md) with explicit placeholders (e.g., {what changed}) and instructions to remove guide comments, which helps separate operational instructions from processed data.
Capability inventory: The skill can execute shell commands, perform network operations via Git/GitHub CLI, and write to the local file system (e.g., /tmp/pr-body.md).
Sanitization: The skill enforces strict validation rules for the PR body (e.g., banning 'TODO' items and ambiguous language) and mandates specific formatting for comments to prevent the inclusion of control sequences, although it relies on the execution environment to ensure shell-safety for interpolated variables.

gh-pr