The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes a workflow for PR reviews that involves fetching remote branches and executing commands within them. This creates a potential surface where malicious instructions embedded in remote repositories could be processed by the agent.
Ingestion points: Remote branch content via git fetch and git worktree add (SKILL.md).
Boundary markers: None; instructions do not include verification steps before execution.
Capability inventory: Shell execution for npm install and npm run dev (SKILL.md).
Sanitization: None detected.
[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands for repository and filesystem management, including destructive operations like rm -rf for removing worktree directories.

git-worktree