GenericAgent

GenericAgent

A self-evolving agent framework that achieves system-level control with just 3.3K lines of seed code.

Agent SkillDevOpen source
Type
Agent Skill
Open source
Yes
GitHub Stars
★ 13.0k
Source
skill-github

Overview

GenericAgent is a minimalist, self-evolving autonomous agent framework. Its core code is approximately 3K lines, powered by nine atomic tools and a ~100-line agent loop, enabling any LLM to gain system-level control over local computers—including browsers, terminals, file systems, keyboard/mouse input, screen vision, and mobile devices (ADB). Each time a new task is solved, the execution path crystallizes into a reusable skill. The longer it's used, the more skills accumulate, forming a personalized skill tree. The framework supports major models like Claude and Gemini, is cross-platform compatible, and features efficient token usage.

Capabilities

  • Self-evolving skill tree
  • System-level control
  • Low token consumption
  • Multi-model support
  • Cross-platform compatibility

Use cases

Automated web browsingCAPTCHA survivalFood delivery order placementStock screening

Setup

Requires: Python 3.11 或 3.12API Key
For detailed installation instructions, refer to [installation.md](docs/installation.md) or [installation_zh.md (Chinese)](docs/installation_zh.md).

This information was compiled by AI from public sources and may contain inaccuracies — please refer to the source.

FAQ

Which models are supported?

Supports major models including Claude, Gemini, Kimi, and MiniMax.

What version of Python is required?

Requires Python 3.11 or 3.12; do not use Python 3.14.

Related skills