The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to interact with the filesystem and Git repositories using the gwt utility and standard git commands to manage task-specific worktrees.
[EXTERNAL_DOWNLOADS]: The agent performs network operations to fetch updates and verify the freshness of remote branches, which involves connecting to external Git hosting services.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external Git repositories, creating a vector for indirect prompt injection.
Ingestion points: The agent reads branch names, commit history, Pull Request URLs, and repository-specific configuration files (SKILL.md).
Boundary markers: The instructions lack explicit requirements for the agent to use boundary markers or to disregard natural language instructions found within Git metadata.
Capability inventory: The agent possesses capabilities for shell command execution (git, gwt) and filesystem modification, which could be leveraged if an injection is successful.
Sanitization: There are no specific requirements for sanitizing or filtering input from repository metadata or external PR links before processing.

operations-worktree