AgentLimb is a browser extension that lets any AI terminal — Claude Code, Cursor, Codex, Windsurf, WorkBuddy, Qwen Office, TRAE Work, Cola, Kimi Work, or any local model — drive every Chromium browser (Chrome, Edge, Brave, Vivaldi) via 16 standardized tools over HTTP, on both Windows and macOS. It uses your existing browser sessions (no headless browsers, no re-login), builds muscle memory to save 85.7% tokens on repeat tasks, and enforces an explicit task lifecycle so silence never equals success.

How does muscle memory save tokens?

The first time AI visits a site, it explores the DOM and records selectors, workflows, and navigation paths as a JSON muscle file. On repeat runs it skips all re-exploration. Real regression data from a Reddit posting task: 23 tool calls → 10, ~12,250 tokens → ~1,750 (85.7% savings), 8–20 min → 30 s–2 min (80–95% time savings).

Yes, AgentLimb is free for personal use under the Business Source License 1.1. Commercial use requires a separate license. The license converts to Apache 2.0 on April 12, 2030.

Is my data private with AgentLimb?

Yes. AgentLimb runs 100% locally. The bridge operates on localhost (127.0.0.1:7791), no data is sent to external servers, and there are no analytics or tracking. Privacy is guaranteed by architecture, not just policy.

Now live on the Chrome Web Store

Stop Watching Your AI
Relearn the Same Task

Record once. Replay forever. 85% fewer tokens.

Works with Claude Code, Cursor, Codex, WorkBuddy, Qwen Office, TRAE Work, Cola, Kimi Work — or any AI terminal, in every Chromium browser (Chrome, Edge, Brave, Vivaldi). No setup, no API keys, your existing sessions.

Install from Chrome Web Store Star on GitHub ⭐️

One click install · Auto-updates · Or grab the latest zip for dev builds

Officially supports Chinese AI terminals

Made for Chinese AI terminals. Ready out of the box.

The same one-prompt onboarding as Claude Code and Codex — copy the prompt, paste it in, approve, done. No adapter, no special config, your existing browser sessions.

TOP 01

Tencent

腾讯 WorkBuddy

Tencent's AI terminal for coding and office work — runs local commands, manages projects, and drives automation end to end.

TOP 02

Alibaba

阿里千问办公

Alibaba's AI office terminal powered by Qwen — documents, data, and web tasks in one workspace.

TOP 03

ByteDance

字节 TRAE Work

ByteDance's AI work terminal — team-grade coding, agent workflows, and enterprise automation.

TOP 04

Cola

A lightweight Chinese AI terminal — quick to start, great for personal automation and everyday tasks.

TOP 05

Moonshot AI

Kimi Work

Moonshot's work terminal with long-context strength — research, reading, and web-heavy tasks.

Plus Claude Code, Codex, Cursor, Trae, Windsurf and any command-capable AI · Works in every Chromium browser: Chrome, Edge, Brave, Vivaldi · Windows & macOS

Works with these AI coding agents

Why AgentLimb

Your AI is wasting tokens doing the same thing every day.

Every task, it re-explores from scratch. AgentLimb fixes that.

💪

90% Fewer Tokens

AI explores a task once and saves the route as a "muscle." Every replay after that skips the exploration entirely — same result, a fraction of the cost.

🔌

Any AI Tool. Any Chromium Browser. Zero Setup.

Claude Code, Cursor, Codex, Trae, Windsurf, WorkBuddy, Qwen Office, TRAE Work, Cola, Kimi Work — paste one prompt and your AI auto-configures in seconds. Works on Windows & macOS, in every Chromium-based browser. No config files, no API keys.

🔑

Your Real Browser. Your Real Sessions.

No headless Chrome. No re-login. No separate instances. Your cookies, logins, and extensions — all intact. If you can see it, your AI can use it.

🔒

100% Local & Private

Everything runs on localhost. No analytics, no tracking, no cloud. Your browsing data never leaves your machine.

🌐

Any Website, Any Platform

Social media, SaaS dashboards, internal tools, email — anywhere your Chrome can go, your AI can operate.

👥

Multi-Account Parallel Control

Run one task across multiple Chrome profiles at once — different accounts, same command. Each profile has explicit identity, can be suspended independently, and receives the full task lifecycle broadcast.

The Core Idea

AI has no memory. Give it one.

Every time your AI posts on Reddit, it rediscovers where the button is. Every time it fills a form, it re-reads the page from scratch. You're paying tokens for work it already did yesterday.

AgentLimb records that route as a "muscle." Next run: one command, zero exploration, the same result — without the re-learning tax.

One exploration per platform. Saved permanently. The more you reuse, the more you save.

        
        First Explore
        Muscle Replay
      

        Tool Calls
        23
        10
      

        Tokens
        ~12,250
        ~1,750
      

        Wall-clock
        8–20 min
        30s–2 min
      
85% Fewer Tokens · 80–95% Less Waiting

Architecture

Three layers. All localhost. Zero cloud.

16 standardized tools, CDP-native. Any AI terminal that can send HTTP can drive your browser.

Your AI Terminal Claude Code / Codex / WorkBuddy / TRAE Work / any agent

HTTP + SSE

AgentLimb Bridge local Node.js · 127.0.0.1:7791

runtime msg

AgentLimb Extension Chrome MV3 · Side Panel UI

Chrome DevTools Protocol

Your Browser logged in · cookies · real sessions

Built For

People who do the same thing every day.

You're not lazy. You just know that repetition is a waste of your time — and your tokens.

📣

The Daily Marketer

Post to 5 platforms without opening a tab. Engage communities, run campaigns — all through your real accounts, zero re-login.

🔍

The Lazy Researcher

Collect data, compare products, track competitors. Your AI runs the route; the muscle handles the next hundred times.

📝

The Repeat Operator

Forms, applications, profile updates, submissions. Record once. Run whenever. No manual steps.

🧪

The QA Engineer

Test real flows on a real browser with real sessions. Save the test path as a muscle — rerun with one command.

Why AgentLimb

Every other approach has a real cost.

Browser automation has been solved — badly — four different ways. Here's the honest breakdown.

	Browser Use / Playwright	BrowseAI / Browserbase	Codex / Claude Computer Use	AgentLimb
Setup	Write scripts, manage dependencies, handle headless mode	SaaS config, per-workflow setup	Mac only (needs desktop env), sandbox required	Copy one prompt. Done.
Element targeting	CSS/XPath — you write and maintain them	Visual AI detection — unstable on updates	Screenshot coordinates — ±1px can miss the target	CDP reads the live DOM — semantic, precise
Token cost per action	None (pure script)	Cloud fees + AI tokens	1,000–3,000 tokens/screenshot × every step	~300 tokens/step. Hot start: 85.7% less.
Repeat task cost	Fixed (script re-runs)	Linear — billed per execution	Linear — re-explores every time, no memory	Decreasing — muscle memory compounds
Login sessions	Extra cookie/session setup	Cloud — can't use your local sessions	OS-level, unaware of browser state	Your real Chrome — already logged in
When the site updates	Selectors break — rewrite scripts	Visual model may degrade silently	Screenshot inference adjusts, but expensively	AI detects mismatch, finds new selector, self-heals muscle
Data privacy	Local ✅	Through third-party servers ❌	Local ✅	100% local — 127.0.0.1 only ✅
AI terminal choice	Any (pure script)	Varies by platform	Bundled with Codex / Claude	Any AI that speaks HTTP
Shared knowledge	Script = locked to one AI	Workflow = locked to platform	No persistent memory	Muscle files = cross-AI, transferable, permanent

Muscle memory is cross-AI: knowledge explored by Claude Code today is reused by Codex tomorrow — same JSON files, same Desktop folder.

Quick Start

Up and running in under a minute.

No account. No API key. No cloud. Install the extension and local runtime, then copy one prompt.

Install the Extension

Install from the Chrome Web Store, or load the extension zip from GitHub Releases.

Start the Bridge

Install the Bridge and CLI package for Windows or run scripts/install.sh on macOS. The Bridge then starts automatically at login.

Copy the Prompt & Paste

Open the side panel, click "Copy Onboard Prompt", paste to Claude Code / Cursor / Codex / any AI terminal. It auto-connects and starts working.

Stop Watching Your AIRelearn the Same Task