AI/TLDR

Open Agents

A reference app for running coding agents in the cloud instead of on your laptop

Autonomous Coding AgentsOpen source
Updated
14 Apr 2026
Language
TypeScript
License
MIT
Coverage
2 stories

What's new

14 Apr 2026

Vercel Labs published Open Agents, an open-source reference app for cloud coding agents with isolated sandboxes, durable workflows and GitHub commit/PR automation.

Latest news

Overview

Open Agents is an open-source reference application for building and running background coding agents. You describe a change in chat, the agent clones the repository into an isolated sandbox, works with file, search, shell and web tools, and can commit, push and open a pull request when it is done — none of which depends on your machine staying awake.

The interesting architectural choice is that the agent runs *outside* the sandbox. The sandbox is only the execution environment, so its lifecycle is managed independently: sessions snapshot and resume, and the model provider can be swapped without touching the sandbox layer. Multi-step execution is durable, built on the Workflow SDK, so a long run survives a restart rather than losing its place.

It is deliberately a template rather than a product: the repository is a Next.js app you deploy yourself and adapt, and there is a Vercel template that stands the whole thing up. Sessions can be shared as read-only links, and voice input is available through ElevenLabs transcription.

What it does

  • Chat-driven coding agent with file, search, shell and web tools
  • Durable multi-step execution on the Workflow SDK, so long runs resume instead of restarting
  • Isolated cloud sandboxes with snapshot-based resume, managed separately from the agent
  • Repository cloning and branch management inside the sandbox
  • Optional auto-commit, auto-push and pull-request creation
  • Read-only session sharing links, and voice input via ElevenLabs transcription

Getting started

The repository is a pnpm monorepo with a Next.js web app; you fill in an env file and run it locally, or deploy it from the Vercel template.

Install dependencies

The project pins its package manager through corepack.

bashbash
corepack enable
pnpm install

Configure the environment

Copy the example env file for the web app and fill in the required values (model provider, sandbox and GitHub credentials).

bashbash
cp apps/web/.env.example apps/web/.env

Run the app

Starts the web UI; from there you point an agent at a repository and give it a task.

bashbash
pnpm web

Or deploy the template

A one-click Vercel template exists at https://vercel.com/templates/template/open-agents if you would rather start from a deployment than a checkout.

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

When to use it

  • Kick off a refactor from your phone and review the pull request when it lands
  • Run several agent sessions in parallel against the same repository, each in its own sandbox
  • Give a team a shared, self-hosted coding agent instead of everyone running one locally
  • Use it as a starting point for a custom agent product rather than building sandboxing and workflow durability yourself

How Open Agents compares

Open Agents alongside other open-source autonomous coding agents tools AI/TLDR tracks, ranked by GitHub stars.

ToolStarsWhat it does
opencode★ 206kOpenCode is an open source AI coding agent that runs in your terminal, with built-in build and plan agents and an optional desktop app.
Claude Code★ 145kAnthropic's agentic coding tool for the terminal, IDE and GitHub: it understands your codebase, executes routine tasks, explains code and handles git workflows from natural-language commands.
OpenAI Codex★ 123kOpenAI's cloud and CLI coding agent that writes features, fixes bugs, and proposes code changes across a repo, running tasks in parallel.
Gemini CLI★ 107kAn open-source command-line AI agent from Google that connects your terminal to Gemini models for reading code, editing files, running shell commands, and searching the web.
Pi★ 104kMinimal terminal coding agent harness with four built-in tools, extended through TypeScript extensions, skills and prompt templates instead of forks.
OpenHands★ 87.2kAn open-source AI software-development agent that plans tasks, edits files, runs commands, and tests code, usable from a terminal CLI, a local web GUI, or a Python SDK.
Cline★ 67.8kOpen-source coding agent that runs as a VS Code or JetBrains extension, a terminal CLI, or an embeddable Node SDK, editing code and running commands with human-in-the-loop approval.
Open Agents★ 5.8kA reference app for running coding agents in the cloud instead of on your laptop