Agents: Add document conversion and pagination to web fetch tool

Agents SDK 0.28.0 adds a beta web_fetch tool for reading URLs. It handles HTML, PDF, and supported Office documents through Workers AI Markdown Conversion, requests Markdown directly when available, and formats JSON for the model.

Long pages come back in smaller windows with offsets to continue reading. Each result includes the final URL, HTTP status, and content type. You control allowed hosts, response size, timeouts, and redirects.

Add the tool to a Pi Durable extension registry before opening the harness:

import { webFetchTool } from "agents/webfetch/pi";

this.registry.install({
	name: "web-tools",
	tools: [webFetchTool({ binding: this.env.AI })],
});

import { webFetchTool } from "agents/webfetch/pi";

this.registry.install({
	name: "web-tools",
	tools: [webFetchTool({ binding: this.env.AI })],
});

The SDK also includes AI SDK and TanStack AI adapters. For another harness, use the public fetchWeb, windowWebFetchPage, and renderWebFetchPage functions to build its tool handler. Direct fetching does not require a Durable Object or browser.

For setup, examples, and request controls, refer to Web fetch.