First-party app
Airtop Airtop

Airtop

Drive real, cloud-hosted browsers with Airtop's AI browser automation: create sessions and windows, load and interact with pages, and extract structured data with natural-language prompts.

stable AI & Machine LearningDeveloper Tools

About

Airtop ships in the w6w first-party pack. It declares 24 actions, 2 health checks, and the host runs its code in a sandbox that never sees the credential.

App id
io.w6w.airtop
Version
0.1.1
Author
w6w
Licence
MIT
Categories
AI & Machine Learning · Developer Tools

Overview

Airtop runs real Chromium browser sessions in the cloud and drives them with an AI layer, so a workflow can automate sites that have no API: log in, click, type, scroll, upload files, and read back what’s on the page — either verbatim or as an answer to a natural-language question.

A session is the cloud browser itself; a window is one tab inside it. Every interaction (click, hover, type, scroll, screenshot, scrape, page query, paginated extraction, summarize, file input) targets a window and is described in plain language — “the login button”, “the search box in the top right” — rather than a CSS selector, so it keeps working when a site’s markup changes. Profiles persist cookies and local storage across sessions (e.g. to stay logged in), and the Files API moves a file in or out of a session’s browser.

This app covers exactly the surface Airtop’s own OpenAPI document (api.airtop.ai/api/openapi.json) marks as its stable, publicly documented API (x-fern-audiences: public). Airtop also exposes a newer act / extract / find-one / find-many / navigate / wait-for-page / llm interface under /v2/sessions/.../windows/..., and an Agents / GTM Engineer / Knowledge Base product surface under /v2/agents, /v2/gtm-engineer, /v2/organizations/.../knowledge-base and related paths — both are left out of this app and explained in the README, rather than guessed at.

Build with Airtop

Three routes to the same 24 actions. The Workflow tab is generated from Airtop's own manifest and carries its real ids, so it is copy-pasteable; the Code and CLI examples are the same call for any action on any app, so every app-specific value in them is a blank you fill in.

Delete File

file-delete

Delete a file.

Get File

file-get

Get a file's metadata and a download URL.

List Files

file-list

List files, optionally filtered by the sessions they're associated with.

Delete Profiles

profile-delete

Delete one or more saved profiles by name.

Create Session

session-create

Start a new cloud browser session.

Get Session

session-get

Get a session's current status and connection details.

List Sessions

session-list

List sessions, optionally filtered by ID or status.

Save Profile on Termination

session-save-profile

Persist this session's cookies and local storage as a named profile when it ends.

Terminate Session

session-terminate

End a session. Already-ended or unknown session IDs are ignored, not an error.

Click

window-click

Click an element on the page, located by a natural-language description.

Close Window

window-close

Close a browser window in a session.

Create Window

window-create

Open a new browser window in a session, optionally loading a URL.

File Input

window-file-input

Attach a previously uploaded file to a file input element on the page.

Get Window

window-get

Get a browser window's current info, including a live view URL.

Hover

window-hover

Hover over an element on the page, located by a natural-language description.

List Windows

window-list

List every browser window open in a session.

Load URL

window-load-url

Navigate an existing browser window to a URL.

Page Query

window-page-query

Ask a natural-language question about the page's content.

Paginated Extraction

window-paginated-extraction

Extract data across multiple pages of results in one call.

Scrape Content

window-scrape-content

Read the page's content verbatim, as text.

Screenshot

window-screenshot

Capture a screenshot of the window's current viewport.

Scroll

window-scroll

Scroll the page — to an element, to an edge, or by a relative amount.

Summarize Content

window-summarize-content

Summarize the page's content.

Type

window-type

Type text into an element, optionally located by a natural-language description.

A workflow step names the app and the action, and the editor fills in the connection when you pick one. This is the Step shape from the workflow spec, carrying Airtop's real ids.

{
  "manifestVersion": "2",
  "name": "airtop-example",
  "steps": [
    {
      "id": "file-get",
      "uses": {
        "app": "io.w6w.airtop",
        "action": "file-get",
        "connection": "conn_YOUR_CONNECTION_ID"
      },
      "with": {
        "fileId": "<fileId>"
      }
    }
  ]
}

Here are some of the things you can do

  • Get File

    read
    file-get
  • List Files

    read
    file-list
  • Create Session

    perform
    session-create
  • Get Session

    read
    session-get
  • List Sessions

    read
    session-list

+19 more actions available

Every app-specific value here is a blank you have to fill in. An app action is reached through the connection that authenticates it, so the address is a connection id, not the app id — and connections belong to your account, so a public page cannot know yours. Create one for Airtop, then fill in the three blanks: conn_YOUR_CONNECTION_ID, the action key, and the parameters that action declares. The call itself is real — the shape is transcribed from the studio's own snippet builder, which prints the same kind of blanks — but nothing in it is specific to Airtop. The Workflow tab is where this app's real ids are.

Install
npm install @w6w/sdk
yarn add @w6w/sdk
pnpm add @w6w/sdk
deno add npm:@w6w/sdk
Code
import { W6wClient, isActionRun } from "@w6w/sdk";

// Reads W6W_BASE_URL and W6W_TOKEN from the environment when omitted.
const client = new W6wClient();

const envelope = await client.run({
  urn: "conn_YOUR_CONNECTION_ID",
  action: "file-get",
  payload: {
    fileId: "<value>",
  },
});

if (isActionRun(envelope)) console.log(envelope.value);
Install the CLI
npm install -g @w6w/cli
CLI
w6w run conn_YOUR_CONNECTION_ID --action file-get --payload '{"fileId":"<value>"}'

Give an AI agent Airtop — without giving it Airtop's credentials. One MCP endpoint exposes every app, function and workflow the caller is entitled to, as tools it can discover and run. Access is granted per team while we onboard.

One tool call
{
  "name": "w6w_invoke",
  "arguments": {
    "ref": "app:io.w6w.airtop#file-get",
    "input": {
      "fileId": "<fileId>"
    }
  }
}

Every tool names its target with a single ref. The app: form above doesn't name a connection at all — the host resolves which of the caller's Airtop connections to sign with, and refuses rather than guesses when the answer is ambiguous.

What the agent gets

Credentials it can't read

The token is attached host-side, at the moment of the call. It is never a tool argument, never in the model's context, and never in a transcript — so a prompt injection has nothing to exfiltrate.

A tool surface scoped to the caller

Tools are derived per end user from what that person has actually connected and is entitled to — not one shared bot identity carrying the union of everyone's access.

A durable workflow in one call

Multi-step work runs on the workflow engine and returns a run handle the agent can poll — retries, branching and state survive the conversation that started them.

Health-aware discovery

Airtop's declared health checks are on the surface too, so an agent can tell "the vendor is down" from "your credential expired" before it burns a retry on either.

The MCP surface is part of the hosted platform. Airtop itself is MIT, and the runtime that executes it is source-available (FSL).

Request MCP access

Health checks

Airtop declares its own checks, so its health is a property of the app rather than something the host guesses at.

service

Airtop platform status

Page-level status from status.airtop.ai (Instatus), enriched with component detail.

quota

Credit headroom