Artificial intelligence news.
Straight from the original announcements.

Cloudflare open-sources Kitesurf browser engine for AI agents

November 13, 2026, 21:20By the AI News Bites Editorial TeamSource: Cloudflare
Cloudflare open-sources Kitesurf browser engine for AI agents
Image: Cloudflare

Cloudflare has open-sourced Kitesurf, a stateless browser for AI agents that runs entirely on Cloudflare Workers. Developers can now inspect the code, run it locally, modify it and deploy their own instance under the Apache 2.0 license.

Key points
  • Kitesurf is now open source under the Apache 2.0 license.
  • The browser engine runs entirely on Cloudflare Workers.
  • It supports CDP clients including Puppeteer.
  • Developers can run Kitesurf locally or deploy it to their Cloudflare accounts.
  • A Workers Paid plan is required for deployment.

A browser engine designed for agents

Kitesurf parses HTML, applies CSS, executes page JavaScript and renders screenshots without relying on a browser container. It supports the Chrome DevTools Protocol, or CDP, allowing clients such as Puppeteer to connect to it.

Cloudflare says its main goal is to provide AI agents with a lightweight browser that can load websites, execute JavaScript and interact with pages while using less CPU and memory than a full browser such as Chromium. The company also says Kitesurf can scale on demand across Cloudflare’s global Workers network.

The release follows Kitesurf’s introduction in August and a September update that added WebMCP support, new APIs and improved compatibility. Cloudflare had previously said it planned to open-source the project so developers could run their own browser engine on their own Cloudflare accounts.

What the open-source repository contains

Kitesurf is built on Cloudflare’s Developer Platform and APIs, together with open-source Rust packages compiled to WebAssembly. Blitz handles HTML parsing, the DOM, layout and painting. Stylo provides CSS parsing, cascading and computed styles, while Boa runs JavaScript inside WebAssembly when JavaScript cannot be run in Dynamic Workers.

Linebender libraries provide text layout and 2D graphics, and Krilla generates PDF output. Cloudflare says these projects made Kitesurf possible, while retaining their respective licenses. The repository’s NOTICE file documents third-party notices.

The project has three main parts: the Engine, which manages browser sessions and client connections; PageScript, which runs pages and their JavaScript; and PageRenderer, which draws the result. The repository also separates the Rust core, DOM shims, renderer, CLI and other packages into folders under its packages directory.

Running and deploying Kitesurf

Developers can run Kitesurf locally by cloning the GitHub repository, installing dependencies with pnpm install --frozen-lockfile and starting the engine with pnpm run dev. The first build compiles the Rust core and its dependencies. Wrangler then starts a local development server at http://127.0.0.1:8787, which can be accessed through Chrome DevTools or programmatically with Puppeteer.

After local testing, developers can deploy the engine to their own Cloudflare account with pnpm run deploy. Cloudflare says a Workers Paid plan is required because Kitesurf uses Dynamic Workers to create isolated Workers for pages. This lets users run Kitesurf themselves instead of using the Browser Run managed APIs provided by Cloudflare.

Cloudflare describes Kitesurf as being under heavy development and is seeking contributions focused on compatibility, reliability and efficiency. The company welcomes code contributions, reproduction pages, automation fixes and documentation corrections, with updates planned through its blog, changelog, Web Platform Tests information and Discord community.

CloudflareKitesurfCloudflare WorkersPuppeteerAI agentsopen-source modelscloud
Source
Cloudflare: Kitesurf is now open source

Related news

AWS shows how to route coding agent requests with NeMo Switchyard
For developersNovember 13, 2026, 23:30

AWS shows how to route coding agent requests with NeMo Switchyard

AWS has published a sample deployment that routes coding agent requests between models on Amazon Bedrock using NVIDIA NeMo Switchyard and LiteLLM. The approach measures…

PegasusAI Studio Connects AI Agents to Real Scientific Computing
Products and toolsNovember 12, 2026, 21:39

PegasusAI Studio Connects AI Agents to Real Scientific Computing

PegasusAI Studio, an open-source platform from USC and research partners, is designed to help AI agents build, run, monitor and debug workflows on real scientific…

NVIDIA Donates Agent Substrate Runtime to CNCF Sandbox
For developersNovember 10, 2026, 16:00

NVIDIA Donates Agent Substrate Runtime to CNCF Sandbox

NVIDIA has donated its open-source Agent Substrate runtime to the Cloud Native Computing Foundation Sandbox, targeting the challenges of running large numbers of…