Posts tagged "Agent-runners"

Agent Runners can now read the comments on pull requests that they created, so you can work with an agent through your normal pull request review process.

When a pull request was opened by Agent Runners, you can follow up with a prompt like “address the comments on the PR.” The agent pulls in the review comments from the linked pull request and uses them as context for its changes. You no longer have to copy feedback from GitHub into your prompt.

This works for any pull request created by Agent Runners. Leave your review on the pull request as usual, then start a new agent run on that task to act on the feedback.

Learn more about Agent Runners.

Google’s Gemini 4 Hydrogen model is now available through Netlify’s AI Gateway and Agent Runners with zero configuration required.

Use the Google GenAI SDK directly in your Netlify Functions without managing API keys or authentication. The AI Gateway handles everything automatically. Here’s an example using the Gemini 4 Hydrogen model:

import { GoogleGenAI } from '@google/genai';

export default async () => {
    const ai = new GoogleGenAI({});

    const response = await ai.models.generateContent({
        model: 'gemini-4-hydrogen',
        contents: 'How can AI improve my coding?'
    });

    return Response.json(response);
};

Gemini 4 Hydrogen is available for all Function types and Agent Runners. You get automatic access to Netlify’s caching, rate limiting, and authentication infrastructure.

Learn more in the AI Gateway documentation and Agent Runners documentation.

Netlify Server lets you run your application’s backend as a long-lived Node.js server. Netlify provisions it, scales it with demand and scales it down to zero when nothing is using it.

Whereas, functions handle one request per instance and finish when they respond, some backends need more. Netlify Server keeps running between requests so that it can:

• Handle many requests at once in a single instance.
• Keep warm database connection pools and in-memory caches.
• Hold WebSocket and Server-Sent Events connections open, for things like live chat and real-time collaboration.
• Run an existing Express, Hono, Fastify or Koa app as-is, without additional adapters.

A server is a single file at netlify/server/index.ts. Here’s a simple way to get started without any framework, with the same signature as a Netlify function:

import type { Context } from '@netlify/server';

export default async (req: Request, context: Context) => {
  return Response.json({ message: `Hello from ${context.geo.city}` });
};

Every deploy, including branch deploys and Deploy Previews, runs its own version of your server. The server can use other platform primitives, such as Netlify Database, Netlify Blobs and the AI Gateway, with no setup.

Full-stack frameworks such as Next.js, Astro and Nuxt don’t run on Netlify Server yet, so for now we recommend using a server in projects that don’t use one of these frameworks.

Early Access

Netlify Server is available on credit-based plans and on Enterprise plans with credits enabled. Its APIs and behavior might change during Early Access.

In Agent Runners, we’re releasing Netlify Server gradually over the coming days, so your agent runs might not use it right away.

Limits

These are the main limits during Early Access:

• A server scales up to 3 instances, or 1 instance on credit-based Free and Personal plans.
• An instance always scales to zero after 5 minutes without requests. After Early Access, you’ll be able to keep a minimum number of instances running, so infrequent requests won’t need to wait for your server to start.

The limits page has the full list.

Pricing

Netlify Server is free to use during Early Access. When it reaches General Availability, usage will be charged in credits based on two meters: reserved memory (in GB-hours) and active CPU time (in vCPU-hours). We’ll announce the cost in credits before then. If your usage during Early Access suggests you’ll need to upgrade your plan or buy more credits, we’ll let you know.

Other metered features in your project are billed as usual, and every request your server handles counts toward the web requests meter. Learn more about pricing.

To get started, read the Netlify Server docs and follow the guide to create your first server.

Netlify Visual Editor is deprecated as of November 10, 2026, and will reach end of life on February 8, 2027. After that date, Visual Editor will no longer be available or supported.

Your sites stay live and your content stays right where it is. The only thing that changes is how you make edits.

Timeline

If your team already has Visual Editor enabled, you can keep using it until February 8, 2027, and the Visual Editor docs still apply until then. No new teams or projects can enable Visual Editor.

How to keep editing

The Visual Editor, previously called “Netlify Create”, was designed before prompting updates was a common way people worked. Depending on how your team works, here’s how to keep editing:

• Using a CMS? Keep editing there. Nothing changes for you.
• Already using Claude or Codex? Point your agent to netlify.ai to get the skills and tools it needs to work with your project.
• Want to edit with a prompt in Netlify? Open your project and tell Agent Runners what to change. There’s nothing to install, configure, or connect first. You’ll see the result in a side-by-side preview, and nothing goes live until you publish.

Edit with Agent Runners

Agent Runners runs on credits, so you need a credit-based plan or an Enterprise plan that uses credits. If you’re on a legacy plan, compare plans or learn how to change your pricing plan.

A few things you can ask Agent Runners for:

• Edit content: “Update the homepage headline and swap the hero image.”
• Fix a layout: “The pricing cards stack badly on mobile. Fix the spacing.”
• Add a page: “Add an About page with our team and a contact form.”

Share feedback and get help

We’ll keep adding to what Agent Runners can do, and we want to hear what you’d like next. Tell us in the Netlify Discord, where the team and other builders are around to help and swap tips.

Learn more in the Agent Runners docs, or reach out to Netlify Support with any questions.

Google’s Gemini 4 Carbon model is now available through Netlify’s AI Gateway and Agent Runners with zero configuration required.

Use the Google GenAI SDK directly in your Netlify Functions without managing API keys or authentication. The AI Gateway handles everything automatically. Here’s an example using the Gemini 4 Carbon model:

import { GoogleGenAI } from '@google/genai';

export default async () => {
    const ai = new GoogleGenAI({});

    const response = await ai.models.generateContent({
        model: 'gemini-4-carbon',
        contents: 'How can AI improve my coding?'
    });

    return Response.json(response);
};

Gemini 4 Carbon is available for all Function types and Agent Runners. You get automatic access to Netlify’s caching, rate limiting, and authentication infrastructure.

Learn more in the AI Gateway documentation and Agent Runners documentation.

You can now power your Netlify Agent Runners runs with Meta’s Muse family of AI models. Choose Muse for your next run and pick the model that fits the task, from Meta’s most capable Muse Wonder 1.0 for complex software engineering and long-horizon agentic work to Muse Spark 1.3, tuned for agentic workflows and coding.

To use a Muse model, open the agent configuration for your run and select Muse as your agent. Then choose a specific model: Muse Wonder 1.0, Muse Spark 1.3, or Muse Spark 1.2. Or you can leave it on Auto to let the agent pick a model for each run. Each model shows its estimated credit cost so you can weigh capability against spend before you start.

Muse joins Claude, Codex, Antigravity, and OpenCode in the agent configuration, so you’re never locked into a single provider and can match the model to the work at hand. You can start using Muse models today with no additional setup if you have a Credit-based pricing plan on Netlify, including our free plan.

Learn more in the Agent Runners documentation.

OpenAI’s GPT-6.1 Terra model is now available through Netlify’s AI Gateway and Agent Runners with zero configuration required.

Use the OpenAI SDK directly in your Netlify Functions without managing API keys or authentication. AI Gateway handles everything automatically. Here’s an example using GPT-6.1 Terra with the Responses API:

import OpenAI from 'openai';

export default async () => {
    const openai = new OpenAI();

    const response = await openai.responses.create({
        model: 'gpt-6.1-terra',
        input: 'Give a concise explanation of how AI works.',
    });

    return Response.json(response);
};

GPT-6.1 Terra is also available across Scheduled Functions, Background Functions, and Edge Functions. You get automatic access to Netlify’s caching, rate limiting, and authentication infrastructure.

Learn more in the AI Gateway documentation and Agent Runners documentation.

MiniMax’s M3.1 Flash model is now available through Netlify’s AI Gateway and Agent Runners with zero configuration required.

Use the OpenRouter SDK directly in your Netlify Functions without managing API keys or authentication. AI Gateway handles everything automatically. Here’s an example using MiniMax M3.1 Flash:

import { OpenRouter } from '@openrouter/sdk';

export default async () => {
  const client = new OpenRouter();

  const response = await client.chat.send({
    chatRequest: {
      model: 'minimax/minimax-m3.1-flash',
      messages: [{ role: 'user', content: 'How can AI improve my coding?' }],
      maxTokens: 400,
    },
  });

  return Response.json(response);
};

MiniMax M3.1 Flash is also available across Scheduled Functions, Background Functions, and Edge Functions. You get automatic access to Netlify’s caching, rate limiting, and authentication infrastructure.

Learn more in the AI Gateway documentation and Agent Runners documentation.

You can now start and manage Agent Runners from your own scripts, tools, and apps through Netlify’s public API.

The new endpoints let you start an agent run, send follow-up prompts, answer an agent’s questions, and review the files it changed. You can also preview the agent run’s work with a deploy URL shared in the API response.

To start a run, choose a project and send a prompt. When you use the [POST /sites/{site_id}/agent_runners](https://open-api.netlify.com/#tag/agentRunner/operation/createSiteAgentRunner) endpoint, your run will start on the production branch by default. To specify a different branch, you can pass a branch or deploy ID. You can also choose the agent, model, and effort, or use ask mode to get answers without changing code.

The public API supports running agents and previewing their work. Opening pull requests, committing changes, and publishing to production aren’t available yet. For progress updates, poll the run and session endpoints.

To get started, explore the Agent Runners endpoints in the Netlify API reference.