NousResearch/hermes-agent
Hermes Agent is a self-improving AI agent by Nous Research designed to learn from experience, create and refine skills, and maintain knowledge across sessions.
Awesome AI Agents › AI-Powered Agents
lsdefine/GenericAgent
GenericAgent is a lean, self-evolving autonomous agent framework designed to enable Large Language Models (LLMs) to achieve comprehensive system-level control over a local computer. With a core codebase of approximately 3,000 lines and a 100-line Agent Loop, it utilizes nine atomic tools to manage browser functions, terminal operations, filesystem interactions, keyboard/mouse input, screen vision, and mobile devices via ADB. A central tenet of GenericAgent is its "evolve skills, don't preload them" philosophy. As the agent successfully completes new tasks, it automatically distills the execution path into reusable "Skills," which are then added to a personal skill tree. This allows the agent's capabilities to continuously expand through usage, moving beyond its initial modest seed code. The project highlights its self-bootstrap proof, claiming that the entire repository's creation, from Git installation to commit messages, was autonomously performed by GenericAgent without human terminal interaction. Key features include its self-evolving mechanism, minimal architecture, robust execution capabilities (e.g., injecting into real browser sessions), high LLM compatibility (supporting Claude, Gemini, Kimi, MiniMax, etc.), and token efficiency, operating with a context window under 30K. It offers various frontends, including a desktop app, a keyboard-driven terminal UI built with Textual, and a Streamlit UI, alongside bot interfaces for platforms like Telegram and WeChat, demonstrating its versatility in deployment and interaction.
https://github.com/lsdefine/GenericAgent
Also in AI-Powered Agents
google-gemini/gemini-cli
Gemini CLI is an open-source AI agent that integrates Google's powerful Gemini model directly into the terminal, enabling developers to perform advanced code understanding, automation, and integration tasks efficiently from the command line.
feder-cr/AIHawk
Open-source AI browser agent that follows natural-language instructions to click, type and read a real browser, usable via MCP or a standalone web UI.
earthtojake/text-to-cad
A library providing agent skills for CAD, robotics, and hardware design, enabling AI agents to generate, inspect, and manage various design artifacts.
browseros-ai/BrowserOS
BrowserOS is an open-source, privacy-first Chromium-based browser that runs AI agents locally to automate browsing tasks and protect user data.
ChatLab/ChatLab
ChatLab is a local-first desktop application that analyzes chat histories from various platforms using AI agents and a SQL engine for insights and visualizations, prioritizing user data privacy.
Integuru-AI/Integuru
Integuru is an AI agent that generates integration code by reverse-engineering platforms' internal APIs using browser network requests and OpenAI models to automate platform interactions without requiring permission.
ev-flow/quark-engine
Quark Agent is an AI-powered tool that enables natural language-based Android APK vulnerability and malware analysis through automated script generation and intuitive workflow design.