SYSTEM Cited by 3 sources

Cloudflare Browser Rendering

Browser Rendering is Cloudflare's managed headless-browser tier: programmatically-controllable Chromium instances running at scale in Cloudflare's edge network, reachable by applications running on Workers and (via proxy) on Containers.

Framework support

Cloudflare ships adapters for the common headless-browser clients:

• Puppeteer
• Playwright
• Stagehand
• Playwright MCP — model-context-protocol wrapper for AI agents driving a browser

Applications talk to Browser Rendering over the Chrome DevTools Protocol (CDP) that Chromium-based browsers already speak.

Node.js compatibility milestone

Cloudflare notes that Playwright on Browser Rendering historically had to rely on memfs to mock a filesystem API, forcing a drift from the upstream Playwright codebase. With Workers' growing Node.js compatibility, node:fs is now natively supported — simplifying the adapter and making Playwright upgrades track upstream.

Seen in

• sources/2026-01-29-cloudflare-moltworker-self-hosted-ai-agent — canonical wiki instance of the CDP-proxy pattern for browser automation. Moltbot expects a local Chromium over CDP; Moltworker instead ships a thin CDP proxy from the Sandbox container back through the Worker to Browser Rendering (via Puppeteer APIs) + a Browser Rendering skill injected into the Moltbot runtime at Sandbox start (patterns/cdp-proxy-for-headless-browser). Demo workloads: Google Maps route + screenshot to Slack, Asian-food search, ffmpeg-generated video from captured browser frames.
• sources/2026-04-15-cloudflare-project-think-building-the-next-generation-of-ai-agents — positioned as Tier 3 of the Project Think execution ladder. Agents escalate to Browser Rendering when a target system has no MCP / API and needs to be driven via "navigate, click, extract, screenshot." Integrated via createBrowserTools(env.BROWSER) in Think's getTools().
• sources/2026-04-16-cloudflare-ai-search-the-search-primitive-for-your-agents — Browser Run (2026-04-16 rebrand of Browser Rendering) is the built-in website crawler inside every AI Search instance that uses a website as its data source. "When using the website as a data source, website crawling using Browser Run (formerly Browser Rendering) is also now a built-in service, meaning that you won't be billed for it separately." Third framing of the same primitive: agent-facing tool (Moltworker CDP proxy) / generated- code-facing capability (Project Think Tier 3) / invisible-crawl- substrate (AI Search).

Related

• systems/cloudflare-workers — the compute tier that drives Browser Rendering.
• systems/chrome-devtools-protocol — the wire protocol Browser Rendering exposes.
• systems/project-think — agent SDK that exposes Browser Rendering as Tier 3 of the execution ladder.
• systems/dynamic-workers — paired in Project Think; the generated TypeScript driving the browser runs in a Dynamic Worker with Browser Rendering as a granted capability.
• concepts/execution-ladder — Tier 3 capability.
• patterns/cdp-proxy-for-headless-browser — the CDP-proxy-from- container-to-managed-browser shape Moltworker realises.
• patterns/additive-capability-ladder — the ladder pattern Browser Rendering slots into.
• companies/cloudflare — operator.