The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various system commands and scripts to manage the Git environment.
Executes git rev-parse --show-toplevel to identify the repository root.
Launches platform-specific scripts (isolate.ps1 or isolate.sh) which then invoke node to run a management script.
[EXTERNAL_DOWNLOADS]: The skill depends on a Node.js script located at ../../worktree-package/scripts/worktree.mjs. While this appears to be a local dependency within a monorepo structure, the file is external to the provided skill package and its content cannot be verified.
[PROMPT_INJECTION]: The skill implements a 'handoff prompt' mechanism in SKILL.md (Step 1 and 3). This is an indirect prompt injection surface where task intent is passed to a new session.
Ingestion points: The handoff-prompt parameter in scripts/isolate.ps1 and scripts/isolate.sh.
Boundary markers: No specific delimiters are used in the scripts for the prompt argument.
Capability inventory: The skill can execute shell commands and launch new Node.js processes.
Sanitization: The instructions explicitly direct the agent to 'Rewrite the request as a compact handoff prompt', which acts as a manual sanitization step by the LLM.

isolate