A2UI
Open protocol for agents to stream declarative UIs without executing arbitrary code.
All ai agents in the directory.
53 listings · 1–20
Open protocol for agents to stream declarative UIs without executing arbitrary code.
Open-source low-code Python framework for agents and YAML cognitive architectures.
Self-hosted Python agent platform with AgentOS runtime and browser control plane.
Open-source React toolkit for production AI chat UIs and copilots.
Microsoft open-source framework for multi-agent chat apps and event-driven agent systems.
Open-source Python and TypeScript framework for building production ai agents
Open-source Python framework and CLI for single- and multi-agent systems with built-in tools.
Open-source library and cloud for ai agents that operate real web browsers.
Durable Objects SDK to build and host stateful ai agents on Cloudflare’s edge.
Framework for automating Android devices with natural-language commands and LLM agents
Open-source tool that uses GPT to migrate codebases between languages or frameworks.
Open-source AI orchestration framework for production RAG pipelines and LLM agents.
Open-source platform for building, debugging, and deploying collaborative GPT agents and workflows.
JavaScript framework for building, visualizing, and managing multi-agent AI workflows.
Enterprise platform for building, deploying, governing, and optimizing ai agents.
LLM observability platform for tracing, evaluating, and monitoring AI applications.
Graph-based framework for building stateful, controllable agent workflows.
Python framework for building multi-agent LLM applications with message-based collaboration.
Platform for observing, evaluating, and deploying ai agents and LLM applications.
Data framework for building LLM applications, RAG systems, and agentic workflows.
No listings match these filters.
Published listings of ai agents, tools, and frameworks. The page defaults to agents; switch kind to Tool or Framework, or open All.
Common industry wording: an agent uses a model to decide next steps, can use tools, and works toward a goal over multiple turns. Assistants are often agent-style products meant to work with you in chat or an app. We use one kind per listing so you can filter; borderline products still get a clear label.
Same catalog chrome. /agents starts on agents; /tools starts with Tool selected — a clearer door if you are searching for an AI tool directory.
Tool: you operate it. Agent: it can take steps on a use case. Framework: you build agents with it. Each listing gets one kind.
Someone submits the product, or we add it from public sources. An editor publishes it. There is no paid boost in the ranking.
They are public pricing buckets on the listing, plus any note on the product page. Plans change; treat the note as a pointer, not a live quote.
Yes. Category hops and the Categories menu open the same list with that category selected so you can scan Coding agent, Marketing, Voice agents, and the rest.