Dev Tools
ai
Vercel's AI SDK: the TypeScript toolkit for building AI apps and agents with any model provider
★ 27.2k starsNOASSERTION
View guide →Open, event-based protocol that connects AI agents to user-facing apps with streaming and shared state.
ag-ui is an open-source dev tools project by ag-ui-protocol (ag-ui-protocol/ag-ui on GitHub). It has 16,391 GitHub stars and 1,492 forks, is released under the MIT license, and was last updated on GitHub on 8 October 2026.
Pick the interaction you need first: streaming chat, shared state between agent and UI, frontend tool calls, or human-in-the-loop approval steps.
Scaffold a new AG-UI application with the create-ag-ui-app tool, or add AG-UI to an existing agent built on a supported framework such as LangGraph or CrewAI.
Connect your agent backend to the frontend over a transport such as SSE or WebSockets, using the reference HTTP implementation and the standard event types.
Open the AG-UI Dojo examples to see each building block in 50-200 lines of code, then adapt one to your own agent and UI.
Agent backends emit events compatible with roughly 16 standard AG-UI event types, so any compliant frontend can render agent runs consistently.
Streaming chat is a core feature, letting users watch tokens and progress as the agent works rather than waiting for a final answer.
Application state is synchronized in both directions, so the agent and the user can both read and update a shared view of the work.
The protocol supports generative UI and structured messages, letting agents render rich components instead of plain text only.
Agents can call tools that live in the frontend, so an agent can trigger actions in the app the user is already looking at.
Built for collaboration patterns where a person reviews, edits or approves agent steps before the agent continues.
Clone the protocol repo to explore the spec, SDKs and the Dojo example apps locally.
git clone https://github.com/ag-ui-protocol/ag-ui.git
cd ag-uiCreate a fresh AG-UI application in seconds with the official scaffolding command from the README.
npx create-ag-ui-app my-agent-appRun or browse the Dojo building-block examples in the repo to see each AG-UI feature in a small, focused app.
cd apps/dojo # see README for the exact commandThe ag-ui-protocol open-source organization maintains it; it grew out of CopilotKit's agent-interactivity work and has first-party and community contributors across many frameworks.
Yes. It is open source under the MIT license, which permits commercial use, modification and distribution.
AG-UI itself is just a protocol and needs no key. The agent you connect behind it will use whatever model provider you choose, which may require your own API key.
They are complementary: MCP gives agents tools, A2A lets agents talk to other agents, and AG-UI brings agents into user-facing applications.
Run npx create-ag-ui-app my-agent-app, or open the AG-UI Dojo to copy a small working example for your framework.
Vercel's AI SDK: the TypeScript toolkit for building AI apps and agents with any model provider
rmcp: the official Rust SDK for the Model Context Protocol — tokio-based servers and clients
The frontend stack for AI agents and generative UI in React, Angular, mobile and chat apps
Lightweight AST-based semantic code search CLI (ccc) that makes coding agents faster and cheaper
Shared memory for coding agents: lessons, repo knowledge, and your preferences, carried across sessions.
Persistent memory for coding agents — frozen, but its core lives on in Caveman.
The most-starred other Dev Tools projects in the LibreAI catalogue — the realistic shortlist if ag-ui is not the right fit.
The open source coding agent
The model-definition framework behind modern AI — thousands of pretrained transformers for text, vision, and audio.
Anthropic's agentic coding tool that lives in your terminal and works across your whole codebase
Cross-platform desktop all-in-one manager for Claude Code, Codex, OpenCode and other coding agents
OpenAI's lightweight coding agent that runs locally in your terminal
Free and open — grab the source, star the repo, and start creating.