AG2 is a GitHub repository maintained by the ag2ai organization for AG2, a project formerly known as AutoGen. AG2 is an open-source AgentOS designed for developing and operating AI agents.
Directory
Search the directory
Search published websites, apps and repositories.
A GitHub repository for generating retro pixel game characters with generative adversarial networks. It includes the TinyHero dataset of 64x64 pixel characters generated from the Universal LPC spritesheet, with 3648 images. Notebooks cover a DCGAN based on the PyTorch tutorial, a conditional DCGAN and a deep convolutional autoencoder. Art is dual licensed under GPL 3.0 and CC-BY-SA 3.0.
Agency is an open-source GitHub repository maintained by neurocult. It is a Go-language library intended to help developers explore large language models and other generative AI tools through a clean, effective, Go-idiomatic interface. The project provides quick start guides, feature documentation, and tutorials for users.
Reliable agent framework built on top of the latest OpenAI Assistants API.
Agent Leaderboard is a Hugging Face Space created by galileo-ai that displays interactive leaderboards ranking large language models on agentic tasks. Users can select a task category, evaluation methodology, and performance metric to generate updated comparison tables and visualizations of model performance.
A standardized way to give AI agents new capabilities and expertise.
Open-source project that supports software development with code generation, analysis, debugging, or documentation.
Uses AI agents to complete tasks, coordinate tools or automate multi-step workflows.
A repository for creating and managing autonomous AI agents with customizable goals and tasks.
An Artificial Intelligence Automation Platform.
agent-opt is an open-source library hosted on GitHub for automated optimization of AI agent workflows. It includes six implemented optimization algorithms, works with any model and any evaluation metric, and is built to support future AGI development loops.
agent-qa is an open-source, self-improving QA agent for software teams. It functions as a test harness with memory, allowing users to write natural language tests for web and mobile applications. The tool learns from each test run, adapts to UI changes, and catches regressions before product release.
Agent2Agent (A2A) Protocol is an open standard that enables secure communication, collaboration, and joint problem-solving between different AI agents. The linked page hosts the official specification documentation for the protocol, covering core operations and enterprise features.
Agent4Rec is a recommender system simulator with 1,000 LLM-empowered generative agents initialized from the MovieLens-1M dataset. Each agent interacts with personalized movie recommendations page by page and performs actions such as watching, rating, evaluating, exiting and interviewing. It accompanies the paper "On Generative Agents in Recommendation."
Agenta is an open-source workspace for AI agents. Agents are built through chat, improved with feedback, and shared with a team, either self-hosted or in the cloud. Integrations and messaging in Slack or Telegram are described on the site.
AgentBench is a multi-dimensional benchmark that consists of 8 distinct environments for evaluating LLMs as autonomous agents in interactive tasks. It assesses reasoning and decision-making abilities in multi-turn open-ended generation, and tests over 25 LLMs including APIs and open-source models.
AgentDock is a writing editor, Chrome extension, open prompt library, and agent API, all stated as live. The editor supports writing, citing, and publishing documents, with citations pulled from Zotero, Mendeley, and DOIs. An AI employee for business use is described as shipping Q3 2026.
AgentField is an open source tool designed to build, run, and scale AI agents structured similarly to APIs and microservices. It supports building production agents using prompts, and its code is hosted in a public GitHub repository.
Uses AI agents to complete tasks, coordinate tools or automate multi-step workflows.
AgentGPT is a browser-based tool for assembling, configuring, and deploying autonomous AI agents. Users create an agent by adding a name and goal, then deploy it. The page labels the project as beta and lists example agents for research, travel planning, and study planning.
Agentic AI is a Hugging Face Space created by user bstraehle that generates text responses to user prompts via the OpenAI API. Users must supply an OpenAI API key, enter a prompt, and select an agent option to receive a text output. The tool uses LangChain and LlamaIndex frameworks.
Interactive Hugging Face app that uses AI agents to complete tasks, coordinate tools or automate multi-step workflows.
Agentic Context Engine (ACE) is an open source tool designed to enable AI agents to learn from experience. It is also offered as a hosted solution at kayba.ai, with supporting resources including proven results documentation, a quick start guide, and runners.
Agentic Engineering Jobs supports AI agents, automated workflows, orchestration, or delegated tasks.