Tabductor FAQs

Question 1

What is Tabductor?

Accepted Answer

Tabductor connects an MCP (Multi-Modal Control Protocol) client to your active Chrome browser tab, allowing AI agents to read page content, interact with elements, and automate tasks directly within your logged-in browser session.

Question 2

How does Tabductor benefit AI agents?

Accepted Answer

It provides AI agents with direct access to your real Chrome session, enabling them to work with sites where login state matters. Agents can perform actions like clicking, typing, navigating, and running custom JavaScript without needing a disposable test browser.

Question 3

Can I use Tabductor with multiple browser tabs simultaneously?

Accepted Answer

Yes, Tabductor is session-oriented. Each connected browser tab is exposed as a distinct session, allowing AI agents to work across multiple tabs in parallel and manage isolated long-running tasks.

Question 4

What kind of page information does Tabductor provide to agents?

Accepted Answer

It provides compact, structured page snapshots with actionable element references. This allows agents to efficiently understand the page state and act on specific elements without processing a full DOM dump, optimizing multi-step workflows.

Question 5

What are the core requirements to use Tabductor?

Accepted Answer

You need Bun, a Chromium-based browser (like Chrome) that can load unpacked extensions, and an MCP client that supports local stdio servers. Setup involves cloning the repo, building it, loading the extension, and configuring your agent to use the MCP server.

Tabductor

Key Features

Use Cases

Tabductor

Key Features

Use Cases