Overview

Discover what Queue AI Agents are, how to spin up your first coding agent, and what you can automate with them.

What is Queue?

Queue is an AI agent platform built for software teams. Each agent runs autonomously in its own environment — reading your codebase, writing code, running tests, and opening pull requests without manual intervention.

Agents are triggered via the API or from your CI pipeline. You define the task — fix this bug, implement this feature, refactor this module — and Queue handles the rest. Every run is logged, auditable, and reversible.

How agents work

When a task is dispatched, Queue creates an isolated sandbox with your repository cloned, dependencies installed, and the agent loaded. The agent reads context, reasons about the problem, writes a solution, and runs your test suite before surfacing a result.

typescript

import { Queue } from ‘@queue/sdk’ const client = new Queue({ apiKey: process.env.QUEUE_API_KEY }) const task = await client.tasks.create({ repo: ‘my-org/my-repo’, prompt: ‘Fix the failing unit tests in src/auth’, branch: ‘queue/fix-auth-tests’, })

Agent types

Queue offers two agent modes: synchronous and queued. Synchronous agents block until complete and return results inline — ideal for short tasks. Queued agents run in the background and emit webhook events as they progress, suited for larger refactors or multi-file changes.

On this page

What is Queue?

Models

Agent types

Synchronous vs. queued

Environment setup

Resources