Browser (Chromium)

Overview

Drives an actual Chromium browser the way a human would — clicking links, typing into search boxes, filling out forms, taking screenshots. Use this when the page is too JavaScript-heavy for the lightweight 'browse' tool, or when the agent needs to interact with the page rather than just read it.

How it works

Powered by Microsoft Playwright over the Chrome DevTools Protocol. Supports three connection modes: launch a fresh Chromium, attach to an existing Chrome session (inheriting all cookies and logins), or use a Playwright-managed persistent profile. Operations include navigate, click, type, fill, fill_form, screenshot (file or base64), scrape, evaluate_js, wait_for, and close. The page state is preserved across tool calls, so multi-step flows like login → search → click result work without re-establishing context. Marked DANGEROUS — agents need explicit permission, and a host allow-list (`allowedHosts`) hard-restricts navigation at the framework layer.

Example

When a user asks:

Log in to my dashboard and download today's report.

the agent calls the tool:

browser(operation="fill_form", fields={"#username": "alice", "#password": "…"}); browser(operation="click", selector="button[type=submit]"); browser(operation="click", selector="a.download-report")

and gets back: the form is filled, the login button is clicked, and the report link is clicked — all in the same persisted browser session.

Configuration

Set these before calling the tool. Values marked required must be present or the tool call will fail.

Use it in a workflow

Wire this tool into a SwarmAI crew. Use the YAML DSL for declarative workflows, or the Java builder API when you want full programmatic control.

YAML DSL

# browser-automation.yaml
name: browser-automation-crew
process: SEQUENTIAL

agents:
  - id: operator
    role: Web UI Operator
    goal: Drive a real browser to perform multi-step user flows
    tools:
      - browser

tasks:
  - id: download-report-task
    agent: operator
    description: Log in to the dashboard, navigate to Reports, and download today's PDF.

Java

import ai.intelliswarm.swarmai.agent.Agent;
import ai.intelliswarm.swarmai.task.Task;
import ai.intelliswarm.swarmai.swarm.Swarm;
import ai.intelliswarm.swarmai.swarm.SwarmOutput;
import ai.intelliswarm.swarmai.process.ProcessType;
import ai.intelliswarm.swarmai.tool.common.BrowserTool;
import org.springframework.ai.chat.client.ChatClient;
import org.springframework.beans.factory.annotation.Autowired;

@Autowired ChatClient chatClient;
@Autowired BrowserTool browserTool;

Agent operator = Agent.builder()
    .role("Web UI Operator")
    .goal("Drive a real browser to perform multi-step user flows")
    .chatClient(chatClient)
    .tool(browserTool)
    .build();

Task downloadTask = Task.builder()
    .description("Log in to the dashboard, navigate to Reports, and download today's PDF.")
    .agent(operator)
    .build();

SwarmOutput result = Swarm.builder()
    .agent(operator)
    .task(downloadTask)
    .process(ProcessType.SEQUENTIAL)
    .build()
    .kickoff();

What it's good for

Real scenarios where agents put this tool to work.

Source

Implementation lives at swarmai-tools/src/main/java/ai/intelliswarm/swarmai/tool/common/BrowserTool.java in the swarm-ai repository.

Open browser on GitHub →