Meet the new Netlify, where developers, agents, and now anyone with an idea build together. See what’s new.

Agent Runners

Dream bigger while your agents work for you.

Agent Runners are cloud coding agents, already connected to your Netlify project and the tools your team uses. Pick your agent, hand it a task, and come back to a live preview.

Start a run from wherever you work

Hand off a task from the dashboard, your terminal, your own tools, or Linear, then get on with your day.

  1. Netlify dashboard

    Start, follow, and review runs from your desk or your phone.

    Build in Netlify
  2. CLI

    Start a run from your terminal with netlify agents:create.

    View the CLI docs
  3. API

    Create and manage runs from your own tools and scripts.

    View the API docs
  4. Linear

    Assign an issue and get a working preview back in the issue.

Netlify CLI ~/my-app
  1. netlify agents:create "Add dark mode to settings"
  2. Run started with Claude
  3. Working in the cloud
  4. Deploy Preview ready

“I can say, ‘Whatever I did last broke the site. Investigate it, fix it, and send me a preview,’ and it does.”

Kick it off. Close your laptop.

Your agent keeps working for minutes or hours, away from your editor. Already use Claude Code, Codex, or Cursor locally? Keep using them, and hand off the work you’d rather not wait on.

It doesn’t stop at “ship”

Agent Runners already know your project and your connected tools. So they don’t stop when the code is written. They build your app, fix it if it fails, and deliver a preview you can see and share.

  • Works with your existing app

  • Checks the changes

  • Gives you a preview

  • Ready for review

Agent run issue #128
  1. Fix the wrong total on the checkout page
  2. Found and fixed the rounding bug
  3. Build failed once, fixed and rebuilt
  4. Deploy Preview ready for review

Work from what your team already knows

Connect your tools once per project. Every run can use them.

Bring your whole team. Keep control.

Decide who can run agents, who can ship, and how much your team spends. Every run is recorded step by step, so you can see what the agent did and review the result before anything goes live.

  1. Choose who runs agents

    Decide who on your team can start runs, across the whole team or project by project.

  2. Let anyone build

    Teammates who don’t code can build with agents too, with the guardrails you set.

    Build with AI in Netlify
  3. Set spending limits

    Cap AI credits for the whole team, or per member. When someone hits their limit, only they’re blocked and everyone else keeps working.

AI enablement acme / team settings
  1. Agent runs: Developers and Builders
  2. Team credit limit set
  3. Member limits set
  4. Builder’s PR awaiting developer approval

Turn your imagination into reality

Drop a file, folder, or .zip to deploy

Frequently asked questions

What people ask before their first run.

What are Agent Runners?

Agent Runners are AI agents that work on your Netlify project in the cloud. Describe a change, and the agent makes it, checks that your app still builds, and gives you a preview to try before anything goes live.

I already use Claude Code, Codex, or Cursor locally. Do I need Agent Runners?

They do different jobs, and most teams use both. Keep your local agent for work you’re steering yourself, connected to Netlify through the CLI or MCP server. Hand a task to Agent Runners when you want it done without you: in the background, in parallel, or while you’re away from your desk.

Do I need to know how to code to use Agent Runners?

No. Describe what you want in plain language, the way you’d explain it to a teammate. If you do code, Agent Runners work alongside GitHub and your local tools.

Which AI agents and models can I use?

Claude, Codex, Gemini, and OpenCode, covering the latest models from Anthropic, OpenAI, and Google, plus open models. There are no keys or extra accounts to set up, and you can switch from one run to the next.

What do I get back when a run finishes?

A working preview of your app with the changes applied. On projects connected to Git, you also get a pull request to review. Projects without Git work too. Nothing goes live until you publish it.

Will a run change my live data?

No. Each run works on its own copy of your data, so your live data stays untouched.

Is my code used to train AI models?

No. Netlify uses third-party providers only for inference, and they aren’t permitted to train their models on anything sent through Netlify. Read about security and privacy for AI features.

How are Agent Runners priced?

Agent Runners are included on Netlify’s Free plan and every paid plan. Runs use credits from your plan, and team owners can set limits for the team and for each member. See pricing.

⌂ 换网址