The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a Python script (scripts/inspect_repository.py) to interface with local Git repositories via subprocess.Popen. The script is designed with security in mind, utilizing argument lists to avoid shell injection and explicitly scrubbing sensitive environment variables such as GIT_CONFIG_PARAMETERS and GIT_DIR to prevent credential exposure.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from Git repositories (e.g., branch names, status output), presenting a theoretical attack surface for indirect prompt injection.
Ingestion points: Data is ingested through Git command output in scripts/inspect_repository.py.
Boundary markers: The SKILL.md instructions mandate that the agent categorize its output into facts, rules, and suggestions, creating a logical boundary for external data.
Capability inventory: The agent can perform branch creation, switching, and renaming operations.
Sanitization: The skill implements validation via git check-ref-format to ensure branch names adhere to safety and project standards before they are acted upon.

git-branch