Let your AI piggyback on your Chrome sessions

What it does

OpenTabs is a local CLI and Chrome extension that lets AI agents invoke web APIs using your existing browser sessions and cookies. Instead of driving the UI with clicks and DOM scraping, it exposes real endpoints—via plugins or built-in network capture—as tools through the Model Context Protocol. If you are logged into Slack or GitHub in Chrome, your AI can call the same APIs the page uses without separate keys or OAuth.

The interesting bit

The project ships roughly 100 plugins exposing around 2,000 tools, and it claims you can point your AI at an arbitrary website to reverse-engineer its API and scaffold a new plugin. That reframes browser automation from “simulate a user” to “reuse the API contract the frontend already has.”

Key highlights

• Runs entirely local with no cloud component, though its anonymous telemetry is opt-out rather than opt-in.
• Defaults to a paranoid security posture: every tool starts disabled, and enabling a plugin requires reviewing its source and setting per-tool permissions (Off, Ask, or Auto).
• Includes built-in browser tools—screenshots, clicking, typing, network capture—for sites that lack a plugin or public API.
• Plugin scaffolding and npm publishing are built in, so adding unsupported services is meant to be straightforward.

Caveats

• Anonymous telemetry is opt-out, not opt-in.
• All third-party integrations are unofficial and explicitly disclaimed, so API changes or policy enforcement are your problem to solve.

Verdict

Worth a look if you’re building AI agents that need to interact with services behind login walls and you’re tired of maintaining brittle browser-automation scripts. Skip it if you need a polished, end-user product or official vendor support.