Autonomous Agents

Browser-focused automation agent built on AutoGen, designed to handle computer actions via AI. Specializes in web-based task automation with autonomous browser control capabilities.

A web-based autonomous AI agent platform that allows users to create custom AI agents with specific objectives. It operates in a loop to generate tasks, execute them, and learn from outcomes, all accessible through a user-friendly browser interface.

A pioneering open-source autonomous AI agent that can perform complex tasks independently using GPT-4. It can create subtasks, launch new agents, write its own code, and recursively debug and improve itself with minimal human intervention.

A pared-down autonomous AI agent (140 lines of code) that uses OpenAI and vector databases to create, prioritize, and execute tasks. It implements a simple task management loop powered by GPT-4 and Pinecone for memory storage.

The simplest self-building autonomous agent that iteratively builds itself by creating and registering tools as required to complete tasks. It dynamically creates Python functions as tools and can handle complex tasks like web scraping and image analysis.

Multi-agent virtual software company framework that uses agents in different roles (CEO, programmer, tester) to autonomously create customized software from natural language ideas. Implements collaborative development processes with specialized agent roles for complete software creation.

An AI-powered productivity copilot that helps users streamline their workflows and manage their time more efficiently. It acts as an autonomous agent for handling complex business and personal tasks.

Autonomous AI software engineer from Cognition Labs that plans, executes, and iterates on software development tasks in a sandboxed workspace with shell, editor, and browser access. Supports human-in-the-loop review and long-horizon work for complex engineering projects.

An open-source research agent designed for software engineering tasks that can understand and modify codebases, run tests, and interact with development environments. Devon represents research into autonomous programming and software development assistance.

AI-driven multi-agent platform designed to automate and streamline software development processes and DevOps tasks. Combines development and operations expertise in autonomous workflows for continuous integration and deployment.

Framework for building collaborative AI systems with automated workflow generation, including case studies for ComfyUI. Features automated agent collaboration patterns for complex creative and technical workflows.

General-purpose AI workspace agent for research, document creation, slides, images, videos, code, and multi-step workflows. Genspark Super Agent combines multiple models and tools to complete complex tasks from prompts.

An autonomous agent designed for comprehensive online research tasks across various domains. GPT Researcher can gather information from multiple sources, synthesize findings, and produce detailed research reports, serving as a research tool for academic and professional investigations.

AI-powered code generation framework that clarifies user requirements and autonomously builds entire software projects based on a prompt. Features interactive requirement gathering and step-by-step code generation for rapid prototyping and development.

A collaborative system comprising a Large Language Model as the central controller and numerous expert models as collaborative executors from Hugging Face Hub. It handles task planning, model selection, task execution, and content generation across dozens of specialized models.

Generalized agent system that leverages Large Language Models for information-seeking and complex task completion. Designed for multi-domain autonomous task execution and knowledge retrieval.

An autonomous AI agent that aims to maximize efficiency by helping users manage their time and tasks more effectively. It focuses on supercharging productivity and making every minute count.

A research-focused codebase agent that runs when new issues and pull requests are created on GitHub. Maige can search entire codebases, run scripts in sandboxes, and write basic code, serving as a research tool for automated software development and code analysis.

General AI agent that executes tasks rather than only answering prompts. Manus uses its own cloud workspace to plan, browse, create files, automate workflows, and deliver completed work products for business and personal tasks.

Intelligent agent for AI engineering and research, integrating with platforms like arXiv and code repositories to assist in planning and execution. Specializes in machine learning engineering workflows and research project automation.

Open-source platform for AI-driven software development agents that can browse, edit files, run commands, and work across codebases. OpenHands is the continuation of OpenDevin under the canonical OpenHands project.

Open-source general-purpose autonomous agent project inspired by Manus. OpenManus provides a configurable agent runtime with browser, file, code, and tool-use capabilities for executing multi-step tasks from natural language.

Open-source operating-system agent for automating desktop, file, web, terminal, and application tasks from natural language. OS-Copilot includes the FRIDAY framework for learning and reusing skills across tasks.

Personal AI assistant that lives in text messaging and lets users ask for tasks to be handled from a familiar contact-style interface. Poke provides recipes and integrations with tools such as Gmail, Google Calendar, Notion, Linear, Asana, GitHub, Todoist, Supabase, Sentry, Vercel, and more.

LLM-based agent framework designed for agentic process automation (APA) in enterprise workflows. Focuses on automating complex business processes through intelligent agent orchestration.

Experimental framework for automating short-form video content creation using AI models. Autonomously generates, edits, and produces video content with minimal human direction.

A dev-first open-source autonomous AI agent framework enabling developers to build, manage, and run useful autonomous agents quickly and reliably. Features include graphical UI, action console, multiple vector DBs, and performance telemetry.

Software engineering agent that takes GitHub issues and attempts to fix them using language models and an agent-computer interface. SWE-agent can also be used for security research and competitive coding workflows.

Open-source AI agent that automates the transformation of GitHub issues into pull requests, streamlining bug fixes and code improvements. Connects issue tracking directly to code changes for accelerated development workflows.

Code-first agent framework from Microsoft designed for data analytics tasks, enabling agents to plan and execute tasks by generating and running Python code. Transforms natural language requests into executable code workflows for data analysis and automation.

Autonomous prediction market agent that publishes live diary and portfolio data while trading and forecasting on Manifold Markets and Moltbook. Exposes an A2A agent card with forecasting, market analysis, and public data endpoints for researchers following long-running autonomous trading experiments.

UI-focused agent from Microsoft designed to interact with and automate tasks within the Windows OS environment. Enables autonomous control of Windows applications through intelligent UI understanding and interaction.

Cutting-edge research project demonstrating an autonomous LLM-driven agent in Minecraft. The first GPT-4-powered embodied lifelong learning agent that continuously explores, writes and executes code to acquire new skills, and self-improves over time without human intervention using automatic curriculum and iterative prompting.

Autonomous Large Language Model agent specialized in solving complex tasks through autonomous planning and tool usage. Features sophisticated task decomposition and execution capabilities for handling multi-step technical challenges.