Jump to content

Connect SuperML | Leeroopedia MCP: Equip your AI agents with best practices, code verification, and debugging knowledge. Powered by Leeroo — building Organizational Superintelligence. Contact us at founders@leeroo.com.

Implementation:Microsoft Playwright PageAgentDispatcher

From Leeroopedia
Revision as of 11:37, 16 February 2026 by Admin (talk | contribs) (Auto-imported from implementations/Microsoft_Playwright_PageAgentDispatcher.md)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Knowledge Sources
Domains Dispatcher, Agent
Last Updated 2026-02-12 00:00 GMT

Overview

Concrete tool for exposing the AI-driven page agent capabilities over the Playwright RPC protocol provided by the Playwright library.

Description

The `PageAgentDispatcher` class extends `Dispatcher` and implements `channels.PageAgentChannel` to bridge the server-side page agent functionality to remote clients. The page agent provides AI-powered automation through three core methods: `perform` (execute a task described in natural language), `expect` (assert an expectation against the page state), and `extract` (extract structured data from the page). Each method delegates to corresponding functions (`pageAgentPerform`, `pageAgentExpect`, `pageAgentExtract`) from the `pageAgent` module, tracks usage statistics (turns, input/output tokens), and maintains a history of operations via a `Context` object. The dispatcher tracks a `Usage` type with token counts.

Usage

Use PageAgentDispatcher when exposing Playwright's AI page agent capabilities over the protocol, enabling natural language interaction with web pages.

Code Reference

Source Location

  • Repository: Microsoft_Playwright
  • File: packages/playwright-core/src/server/dispatchers/pageAgentDispatcher.ts

Signature

export class PageAgentDispatcher extends Dispatcher<SdkObject, channels.PageAgentChannel, DispatcherScope> implements channels.PageAgentChannel {
  _type_PageAgent: boolean;
  _type_EventTarget: boolean;
  constructor(scope: PageDispatcher, options: channels.PageAgentParams);
  async perform(params: channels.PageAgentPerformParams, progress: Progress): Promise<channels.PageAgentPerformResult>;
  async expect(params: channels.PageAgentExpectParams, progress: Progress): Promise<channels.PageAgentExpectResult>;
  async extract(params: channels.PageAgentExtractParams, progress: Progress): Promise<channels.PageAgentExtractResult>;
}

Import

import { PageAgentDispatcher } from '../server/dispatchers/pageAgentDispatcher';

I/O Contract

Inputs

Name Type Required Description
scope PageDispatcher Yes Parent page dispatcher scope
options channels.PageAgentParams Yes Agent configuration options
params.task string Yes (perform) Natural language task description
params.expectation string Yes (expect) Natural language expectation to verify
params.instruction string Yes (extract) Natural language extraction instruction

Outputs

Name Type Description
usage Usage Token usage statistics (turns, inputTokens, outputTokens)
data string Extracted data (for extract method)

Usage Examples

import { PageAgentDispatcher } from '../server/dispatchers/pageAgentDispatcher';

const agent = new PageAgentDispatcher(pageScope, agentOptions);
const { usage } = await agent.perform({ task: 'Click the login button' }, progress);
const { usage: expectUsage } = await agent.expect({ expectation: 'The page shows a welcome message' }, progress);
const { data } = await agent.extract({ instruction: 'Get all product names' }, progress);

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment