The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs Git repository management by executing a local Node.js script (worktree.mjs) through PowerShell and Bash wrapper scripts. These wrappers use secure argument passing to prevent command injection.
[DYNAMIC_EXECUTION]: The skill documentation describes a workflow that generates temporary cleanup scripts (.ps1 or .sh). These are intended for local environment maintenance and are presented to the user for manual execution.
[PROMPT_INJECTION]: The skill includes a parameter for user intent (UserInstruction), which is passed to the underlying integration tool. This presents a surface for indirect prompt injection if the intent is sourced from untrusted external data, though it is handled as a literal string argument in the execution wrappers.
Ingestion points: UserInstruction parameter in scripts/integrate.ps1 and instruction in scripts/integrate.sh.
Boundary markers: No specific delimiters or safety instructions are applied to the intent string within the shell wrappers.
Capability inventory: Git operations (merging, branch/folder deletion) and script generation.
Sanitization: The wrapper scripts do not sanitize the intent string, relying on the target Node.js script for processing.

integrate