Skip to main content
PAS
Available now

Controlled web browsing and automation with strict allowlists

Headless browsing for research and data collection. Domain allowlists and rate limits prevent abuse.

See research demoView browser policies

What Browser & RPA does

Browser & RPA provides controlled web browsing and automation capabilities. Navigate pages, extract content, capture screenshots, and perform simple form fills—all within strict domain allowlists and rate limits to prevent abuse.

Reader capabilities

  • Navigate to allowlisted domains
  • Extract page content and metadata
  • Capture screenshots
  • Wait for page conditions

Writer capabilities Gated

  • Fill forms (requires approval)
  • Click buttons and links (gated)
  • Download artifacts (gated)
Off by default: Browsing requires explicit domain allowlisting. Outbound connections are gated to prevent abuse.

Who benefits from Browser & RPA

Individuals

Research tasks with citations and evidence

Example: "Collect sources on topic X" — navigate, extract, and compile citations with screenshots.

Teams & Managers

Repeatable intel gathering or form workflows

Example: Automate weekly competitor research with allowlisted domains and rate limits.

Developers & IT

Deterministic runs with artifact capture

Example: Integrate browser automation into CI/CD with predictable schemas and outputs.

Security & Compliance

Domain allowlists and outbound gating

Control: Require explicit domain approval. All navigation and actions logged.

How it works

1

Navigate to allowlisted domain

Use browser.goto with a URL. Orchestrator checks domain allowlist before proceeding.

2

Wait for page conditions

Wait for elements, network idle, or custom conditions before extracting content.

3

Extract content or capture screenshot

Use browser.query to extract text or browser.screenshot to capture evidence.

4

Optional: Fill forms or click (gated)

Use browser.fill_form to interact with pages. Requires approval.

Safety & approvals: Domain allowlist enforced. Rate limits prevent abuse. All navigation logged.

Example workflows

Collect sources and compile citations

Allowlisted domains
Input:

"Collect sources on AI safety from allowlisted sites"

Steps:
  1. browser.goto (navigate to each allowlisted domain)
  2. browser.query (extract article titles and excerpts)
  3. browser.screenshot (capture evidence)
  4. Compile citations with links and screenshots
Output:

Citation table with sources, quotes, and visual evidence

Fill a simple web form

Approval before submit
Input:

"Fill the contact form on example.com"

Steps:
  1. browser.goto (navigate to form page)
  2. browser.fill_form (populate fields) — pauses for approval
  3. browser.screenshot (capture confirmation)
Output:

Form submitted with confirmation screenshot and timestamp

Weekly competitor research

Allowlisted domains
Input:

Automate weekly competitor intel gathering

Steps:
  1. browser.goto (navigate to competitor sites)
  2. browser.query (extract pricing, features, updates)
  3. browser.screenshot (capture key pages)
  4. Compile report with changes from last week
Output:

Weekly intel report with screenshots and change highlights

Technical details

Key tools

  • browser.goto
  • browser.query
  • browser.screenshot
  • browser.fill_form (gated)
  • browser.download (gated)
View tool schemas

Configuration

  • DOMAIN_ALLOWLIST — allowed domains (required)
  • CONCURRENCY — max parallel browsers (default: 2)
  • TIMEOUTS — navigation and wait timeouts
  • RATE_LIMIT — requests per minute per domain

Performance notes

  • Navigation: 1-5s depending on page complexity
  • Query: 100-500ms per selector
  • Screenshot: 200-1000ms depending on viewport size

Observability

  • Navigation counts and blocked domains
  • Query latency and error rates
  • Form fill approvals and rejections

Security posture

Off by default

Browsing requires explicit enablement and domain allowlisting.

Domain allowlists

Only allowlisted domains can be navigated. Attempts to visit other domains are blocked.

Rate limits

Requests per minute per domain are capped to prevent abuse.

Audit logs

All navigation, queries, and form fills logged with timestamps and approval decisions.

Roadmap & status

Available

Current features

  • Navigate, extract, and screenshot
  • Domain allowlists and rate limits
  • Gated form fills
Planned

Coming soon

  • Richer RPA primitives (drag, hover, etc.)
  • Anti-automation safeguards (CAPTCHA detection)
  • Session persistence and cookies
View full roadmap

Frequently asked questions

Ready to automate web research safely?

Get started with Browser & RPA in minutes. Domain allowlists and rate limits keep you safe.

See research demoView examplesTalk to us