Skip to main content
📖 The AI Tool Bible
GenericAgent logo

GenericAgent

Not yet source-checked

Self-evolving agent: grows skill tree from 3.3K-line seed, achieving full system control with 6x less token consumption

Machine-drafted from the vendor's site · not yet source-checkedProfile updated Link checked How we rate

Free· MIT licenseAgents

In short

GenericAgent is a free, self-evolving agent that grants LLMs system control with high token efficiency. It grows a personal skill tree from a minimal codebase to automate local tasks.

Best for

Developers and power users seeking a lightweight, self-improving autonomous agent that can control local systems and browsers with high token efficiency.

Skip if

Users who require pre-loaded, out-of-the-box specialized skills without the need for the agent to evolve them through use, or those using Python 3.14.

GenericAgent is a minimal, self-evolving autonomous agent framework designed to grant any LLM system-level control over a local computer. The core consists of approximately 3,000 lines of code and a 100-line agent loop, utilizing nine atomic tools to interact with the browser, terminal, filesystem, keyboard/mouse, screen vision, and mobile devices via ADB. Its primary feature is self-evolution: as it solves new tasks, it automatically crystallizes execution paths into reusable skills, forming a personal skill tree that grows over time. The framework is token-efficient, operating with a context window under 30K tokens, which is significantly less than other agents. It supports major models like Claude, Gemini, Kimi, and MiniMax, and is distributed under the MIT license. The project includes frontends for terminal, web, and various IM platforms like Telegram and Discord.

Editor's take

The page documents GenericAgent as an open-source, MIT-licensed autonomous agent framework that emphasizes minimal code size and self-evolving capabilities. It details a layered memory system and a set of nine atomic tools for system control. The documentation highlights token efficiency and the ability to run on real browser sessions to avoid bot detection. The page is silent on specific enterprise support tiers or commercial licensing options beyond the mention of a single authorized commercial partner, DintalClaw.

— The AI Tool Bible editorial team

Pros

  • ✅ Self-evolving: automatically crystallizes each task into a Skill
  • ✅ Minimal Architecture: ~3K lines of core code with no complex dependencies
  • ✅ Token Efficient: <30K context window, a fraction of other agents
  • ✅ Strong Execution: TMWebdriver injects into a real browser preserving login sessions
  • ✅ High Compatibility: Supports Claude, Gemini, Kimi, MiniMax and is cross-platform

Cons

  • ⚠️ Requires Python 3.11 or 3.12; Python 3.14 is incompatible with dependencies
  • ⚠️ Pre-installed SOPs are written in Chinese, though GA can rewrite them
  • ⚠️ Advanced capabilities require instructing the agent to set them up rather than pre-installing

Facts

Platforms
cli, web, windows, macos, linux, chrome-extension
API available
Yes
Open source
Yes (license: MIT)
GitHub stars
14,409 (checked Nov 12, 2026)
Last GitHub push
Nov 6, 2026
First commit
Jan 2026

Where this tool sits

Use cases

Food delivery ordering via app navigationQuantitative stock screeningExpense tracking via ADBBatch messaging via WeChatCAPTCHA survival in real browser sessions

Frequently asked

How much does GenericAgent cost?
GenericAgent is completely free. It is distributed under the MIT license, allowing developers and power users to access its self-evolving capabilities without any subscription fees or usage charges.
Which AI models are compatible with GenericAgent?
The framework supports major models including Claude, Gemini, Kimi, and MiniMax. It is designed to work with these LLMs to provide system-level control over local computers, browsers, and mobile devices via ADB.
Is GenericAgent difficult to set up or learn?
It is designed for developers and power users. The core consists of approximately 3,000 lines of code and a 100-line agent loop. While it requires the agent to evolve skills through use, it is built for high token efficiency and lightweight operation.
What platforms and interfaces does GenericAgent support?
The project includes frontends for terminal, web, and various IM platforms like Telegram and Discord. It can interact with browsers, terminals, filesystems, keyboard/mouse, screen vision, and mobile devices via ADB.
How does GenericAgent handle token consumption?
It is highly token-efficient, operating with a context window under 30K tokens. This represents a 6x reduction in token consumption compared to other agents, making it suitable for users seeking lightweight, self-improving autonomous agents.
See alternatives to GenericAgent →

Explore related

Reviews