What is BabelWrap?

The web, as an API, for your agents.

BabelWrap is a lightweight HTTP API and MCP server that lets AI agents interact with any website — navigate, read, click, fill forms, submit, and extract data — without spinning up a full browser stack.

Your agents think in actions, not DOM selectors. Instead of writing fragile CSS selectors or XPath expressions, agents describe what they want in plain English: "click the Login button", "fill the email field", "extract all product prices."

BabelWrap translates these natural-language instructions into real browser interactions and returns a structured, LLM-readable snapshot of the page after every action.

Architecture

AgentClaude / GPT / LangChain / custom
MCP Server18 tools, stdio transport
REST APIFastAPI backend
Session ManagerCookies, auth, state
Playwright EngineHeadless Chromium
Target Website
ComponentRole
BabelWrap ProxyConverts any webpage into a structured, LLM-readable snapshot
Playwright EngineExecutes actual browser interactions headlessly
Session ManagerMaintains state (cookies, auth, page position) across actions
REST APIWhat developers integrate directly via HTTP
MCP ServerWraps the REST API so agents can use it natively as MCP tools

For a deeper look at the internals, see Architecture.

How is BabelWrap different?

FeatureBrowserbase / Steel / Playwright CloudBabelWrap
Interface Raw browser — agents must know CSS selectors, XPath Semantic interface — agents describe targets in natural language
Resource usage Full Chromium instance per session Shared browser pool, lightweight session state
Agent integration Custom glue code required First-class MCP server; drop-in tool for Claude, GPT, LangChain
Output format Raw HTML / DOM Structured snapshot: inputs, actions, forms, navigation, alerts
Cost Per-minute browser billing Per-action billing starting at $0/month
Reliability Agents break when DOM changes LLM-based element resolution adapts to layout changes

The Snapshot Format

Every action returns a structured snapshot that an LLM can immediately understand and reason about. This is BabelWrap's core innovation.

Text representation (what an agent sees):

URL: https://example.com/login
TITLE: Sign In -- Example

CONTENT:
  Welcome back. Sign in to continue.

INPUTS:
  [email-field] Email address (text, empty)
  [password-field] Password (password, empty)
  [remember-me] Remember me (checkbox, unchecked)

ACTIONS:
  [sign-in-btn] Sign In (button, primary)
  [forgot-password] Forgot password? (link -> /reset)
  [create-account] Create account (link -> /signup)

NAVIGATION:
  Home | Products | Pricing | Blog | Docs

ALERTS:
  (none)

The snapshot includes:

  • url / title — current page location and document title
  • content — main readable text on the page (up to 15,000 characters)
  • inputs — all form fields with their current values
  • actions — all clickable elements (buttons, links)
  • navigation — site navigation links
  • alerts — any visible error messages, success banners, or warnings
  • forms — logical groupings of fields and their submit buttons
  • tables — structured table data with headers and rows
  • lists — ordered and unordered list items
  • frames — inputs and actions detected inside same-origin iframes

For the full JSON format and field reference, see The Snapshot Format.

Pricing

TierPriceActionsConcurrent Sessions
Free$0500 / month2
Usage-based$0.01 per actionUnlimited50

That's $10 per 1,000 actions. No monthly subscription, no commitment — you pay only for what you use. Free tier resets on the first of each month.

Site Mapping: $10 per site to generate typed tools your agent can call directly (e.g., linkedin_search_jobs). Charged only if mapping succeeds. Pre-mapped sites from the public catalog are free to use. Full pricing details →

Get Started