The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard development tools and version control commands, including git, gh (GitHub CLI), and various test runners (npm, pytest, cargo, go). These operations are essential to the skill's primary purpose of automating development workflows.
[INDIRECT_PROMPT_INJECTION]: The skill processes repository-derived data such as branch names and commit messages, which are interpolated into shell commands. While this presents a potential attack surface, the instructions are well-structured and focused on legitimate development tasks.
Ingestion points: The skill reads branch names (<feature-branch>), base branch names, and commit history from the local repository state.
Boundary markers: It uses clear delimiters for PR templates (e.g., EOF here-docs) to prevent content confusion during PR creation.
Capability inventory: The skill utilizes git for branch management, gh for remote PR creation, and local test runners for verification.
Sanitization: The instructions rely on standard command-line utility behavior for handling arguments, with no specific mention of manual sanitization for branch names or titles.

finishing-a-development-branch