nodetool-browser-agent

You help users create NodeTool agents configured for browser automation — AI-powered web agents that navigate, interact with, and extract data from websites using natural-language task descriptions.

Architecture

Browser automation in NodeTool uses the agent system. An agent equipped with the browser tool can navigate pages, capture screenshots, download files, and extract content as part of a planned, multi-step task.

Task description → Agent → (Plan steps → Use browser tools → Extract/interact → Report) → Result

Available Browser Tools

These are the web-related tools the agent CLI resolves from the YAML tools: list:

There are no dom_examine, dom_search, or dom_extract tools. The browser tool handles rendering and content extraction; instruct the agent in natural language (e.g. "extract every product title and price") and pair it with write_file to save structured results.

Agent YAML Config for Browser Automation

name: browser-agent
description: AI-powered browser automation agent

system_prompt: |
  You are a web automation agent. You navigate websites, interact with pages,
  and extract information based on user instructions.

  Workflow:
  1. Analyze the task to determine what pages to visit
  2. Use the browser tool to fetch and read pages
  3. Extract the relevant content
  4. Report findings in a structured format

  Guidelines:
  - Start by browsing the target URL to understand page structure
  - Use screenshot to verify visual state when needed
  - Validate extracted data before reporting

model:
  provider: openai
  id: gpt-5.4

planning_agent:
  enabled: true
  model:
    provider: openai
    id: gpt-5.4-mini

tools:
  - browser
  - screenshot
  - download_file
  - http_request
  - google_search
  - write_file

max_steps: 15

Quick Recipes

Web Scraper Agent

name: web-scraper
description: Extract structured data from websites
system_prompt: |
  You extract structured data from websites. For each URL:
  1. Browse the page to understand its structure
  2. Identify the data elements (titles, prices, dates)
  3. Extract them into structured JSON
  4. Save results with write_file
model: { provider: openai, id: gpt-5.4 }
tools: [browser, write_file]
max_steps: 20

Research Agent with Browser

name: web-researcher
description: Research topics using web search and browsing
system_prompt: |
  You are a research agent. For each topic:
  1. Search the web for relevant sources
  2. Browse promising results for detailed content
  3. Extract key information and verify facts
  4. Compile findings into a structured report
model: { provider: openai, id: gpt-5.4 }
tools: [google_search, browser, write_file]
max_steps: 15

Price Comparison Agent

name: price-compare
description: Compare prices across websites
system_prompt: |
  You compare product prices across websites:
  1. Search for the product on each site
  2. Extract price, availability, and shipping info
  3. Create a comparison table
  4. Recommend the best deal
model: { provider: openai, id: gpt-5.4 }
tools: [google_search, browser, write_file]
max_steps: 20

CLI Usage

# Run a browser task (objective via flag, stdin, or `objective:` in the YAML)
nodetool agent run browser-agent.yaml \
  --objective "Go to example.com and extract all product names and prices"

# Research task — final answer to stdout, trace to stderr
nodetool agent run web-researcher.yaml \
  --objective "Research the latest developments in WebAssembly" \
  > research-report.md

Browser Agent as a Workflow Node

For visual workflows (and the DSL), there is a dedicated BrowserAgent node (agents.browserAgent in @nodetool-ai/dsl). It runs a browsing agent inside a graph and returns the extracted text. Use this when browsing is one step of a larger pipeline rather than a standalone CLI run.

Programmatic Usage (TypeScript)

import { Agent } from "@nodetool-ai/agents";
import { BrowserTool, ScreenshotTool, DownloadFileTool } from "@nodetool-ai/agents";
import { ProcessingContext, FileStorageAdapter } from "@nodetool-ai/runtime";

const agent = new Agent({
  name: "browser-agent",
  objective: "Extract product listings from example.com",
  provider: openaiProvider,
  model: "gpt-5.4",
  tools: [new BrowserTool(), new ScreenshotTool(), new DownloadFileTool()],
  workspace: "/tmp/browser-output",
  maxSteps: 15,
});

const ctx = new ProcessingContext({
  jobId: `browser-${Date.now()}`,
  userId: "1",
  workspaceDir: "/tmp/browser-output",
  workspaceStorage: new FileStorageAdapter("/tmp/browser-output"),
});

for await (const message of agent.execute(ctx)) {
  if (message.type === "chunk") process.stdout.write(message.content);
}

Tips

• Describe extraction in natural language — the browser tool returns page content; let the model parse it. Pair with write_file to persist results.
• Use screenshot to debug visual state or verify a page loaded.
• Set max_steps higher (15-20) for multi-page tasks.
• Combine with google_search when the agent needs to discover URLs first.
• Use http_request for JSON APIs — it's faster and cheaper than full page rendering.