Parallel Agents

Verun's core feature is running multiple AI agents at the same time. Each task gets its own isolated git worktree so agents never interfere with each other or with your main branch.

How Worktree Isolation Works

Every task creates a separate git worktree — a full checkout of your repository on a dedicated branch. Agents read and write files only inside their own worktree. Your main branch and all other tasks remain completely untouched.

your-project/
├── .git/
├── .verun/
│   └── worktrees/
│       ├── fizzbuzz-dragon-472/     ← Task 1
│       ├── nullpointer-phoenix-891/ ← Task 2
│       └── helloworld-falcon-305/   ← Task 3
├── src/
└── package.json

Running Multiple Tasks

1. Create the first task — select a base branch and agent.
2. Create a second task with a different goal.
3. Both agents work in parallel, each in their own worktree.

There is no hard limit on concurrent tasks. Practical limits depend on your machine's CPU, RAM, and disk I/O.

Port Assignment

Verun automatically allocates 10 unique ports to every task — no configuration needed. The ports are injected as environment variables into all hooks and the start command:

Use $VERUN_PORT_0 as your app's port in the start command:

PORT=$VERUN_PORT_0 pnpm dev

Each task's dev server runs on a different port, so you can have every task running at once without conflicts.

Multiple Sessions Per Task

You can run multiple agent sessions within the same task — each session shares the same worktree and branch. This lets you compare approaches or continue work with a different agent or model without switching tasks.

Merging Parallel Work

When a task is ready:

1. Review the diff in Source Control (right panel).
2. Commit the changes from the Git actions bar.
3. Push the branch and create a PR — all from within Verun.
4. Merge via GitHub (or your git host).
5. Archive the task — Verun removes the worktree.

Other parallel tasks continue unaffected while you merge.