The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of local shell scripts and system tools to manage parallel agent sessions.
Executes node scripts/orchestrate-worktrees.js which processes orchestration plans.
Utilizes system-level commands including tmux, git worktree, and dmux for environment management.
Invokes agent harnesses like claude, codex, and dmux within isolated panes.
[PROMPT_INJECTION]: There is a potential for indirect prompt injection via the plan.json configuration format used by the orchestration helper.
Ingestion points: The plan.json file is read and parsed by scripts/orchestrate-worktrees.js.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the documentation for the JSON schema.
Capability inventory: The script has the capability to execute arbitrary shell commands (via the launcherCommand property), create git worktrees, and write task-related files to the file system.
Sanitization: The documentation does not describe any sanitization or validation logic for the launcherCommand or workers task strings before they are executed in a shell context.

dmux-workflows