What problem does it solve?
Worktrunk addresses the complexity of running multiple AI agents in parallel without colliding changes. It provides a unified CLI to manage Git worktrees as isolated environments for each agent, enabling scalable experimentation and lifecycle management.
Core Features & Use Cases
- Isolated workspaces per agent: Each agent runs in its own worktree to avoid cross-editing.
- Lifecycle hooks: pre-start, post-start, pre-merge, pre-remove, and more trigger automation around worktree lifecycles.
- Deterministic port mapping: Use hash_port and templates to allocate stable ports per worktree.
- Template-driven automation: Variables, filters, and functions enable dynamic, repeatable workflows.
- Security and approvals: Hook approvals and guardrails for project hooks.
Quick Start
Install Worktrunk, set up shell integration, and create a new worktree for an agent with wt switch --create feature-branch and run it.