The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard system commands for git operations, including staging files (git add), creating commits (git commit), and pushing to remote repositories (git push). The commands are well-structured and use standard shell variable assignment for the current branch, which is a safe practice.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from local file differences and user-provided descriptions to generate commit messages. The mandatory evidence chain is as follows:
- Ingestion points: Output from git status and git diff in SKILL.md is read into the agent's context.
- Boundary markers: No explicit boundary markers or 'ignore' instructions are provided when processing diff outputs.
- Capability inventory: The skill has the ability to read the file system (git diff), write to the repository (git commit), and perform network operations (git push).
- Sanitization: The instructions guide the agent to use specific templates (Conventional Commits) and quoted HEREDOCs (cat <<'EOF') in references/coauthor-attribution.md, which helps prevent simple command injection but does not eliminate indirect prompt injection risk from processed content.

commit-and-push