The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses git and gh (GitHub CLI) to interact with repositories, which is the primary purpose of the skill. Commands include checking authentication status, managing branches, committing staged changes, and creating draft pull requests. All commands are standard for PR management.
[INDIRECT_PROMPT_INJECTION]: The skill processes external, potentially untrusted data to generate its output, which represents an attack surface for indirect prompt injection.
Ingestion points: The skill reads git diff outputs, commit messages, and project specification files (e.g., state.md, spec.md, design.md) as described in references/open.md Step 1.
Boundary markers: references/writing-the-body.md establishes clear behavioral constraints for the agent, such as forbidding confidence claims ("Never claim confidence") and inventorying tests.
Capability inventory: The skill has the capability to execute shell commands for repository management, including git push and gh pr create.
Sanitization: While no technical sanitization is applied to the input data, the skill mandates a "scan test" and verification against thirteen writing patterns to ensure the generated body remains relevant and scannable.

pr