CopilotKit

CopilotKit is an open-source TypeScript/React SDK for building agent-native applications, generative UIs, and AI chat interfaces. Developed by Tawkit, Inc., it sits at the intersection of frontend development and agentic AI, providing the tooling to connect any AI agent backend to a user-facing application. The project is published under the MIT License and has accumulated over 31,000 GitHub stars as of mid-2026.

What It Is

CopilotKit is a fullstack agentic frontend framework that lets developers embed AI agents directly inside their web applications. Rather than building a standalone chatbot, developers use CopilotKit to wire agents into existing React or Angular apps so agents can render UI components, share application state, and pause for human input — all within the same product surface. The company describes it as "the enterprise agentic frontend stack."

The AG-UI Protocol

CopilotKit is the company behind the AG-UI (Agent–User Interaction) open protocol, which defines a bi-directional, real-time event stream between agentic backends and user-facing frontends. The protocol uses 16 real-time event types and is designed to replace the M×N integration problem (M agent frameworks × N frontend clients) with an M+N model. According to the CopilotKit website, AG-UI has been adopted by Google, LangChain, AWS, Microsoft, Mastra, PydanticAI, and others as first-party integrations.

Core Feature Set

The SDK ships with several distinct capabilities:

• Generative UI — Agents dynamically render developer-defined React components at runtime based on user intent and agent state, using the useCopilotAction and useComponent hooks.
• Shared State — A synchronized state layer that both agents and UI components can read from and write to in real time via the useAgent hook.
• Human-in-the-Loop — Agents can pause execution to request user confirmation, input, or edits before continuing a workflow.
• Chat UI Components — Pre-built, customizable chat interfaces (CopilotSidebar, etc.) with streaming support, or fully headless mode for custom builds.
• Threads + Persistence — Automatic conversation threading and cross-session state persistence.
• Backend Tool Rendering — Agents call backend tools that return UI components rendered directly in the client.

Developer Setup Path

CopilotKit is designed to be added to new or existing projects with a single CLI command:

npx copilotkit@latest create

or for existing apps:

npx copilotkit@latest init

This installs core packages, configures the provider, connects agent-to-UI streaming, and prepares the app for deployment. The runtime supports LangGraph, CrewAI, Mastra, Agno, LlamaIndex, AG2, AWS Strands, Google ADK, Microsoft Agent Framework, Pydantic AI, and Vercel AI SDK as agentic backends.

Update: v1.59.3

The latest release is v1.59.3, published on June 3, 2026. The repository is actively maintained with frequent releases — the GitHub project shows 31,988 stars, 4,126 forks, and a last push date of June 4, 2026. The product direction is moving toward the Enterprise Intelligence Platform, which adds analytics, self-learning (CLHF — Continuous Learning from Human Feedback), VPC/on-prem deployment, and multimodal storage on top of the open-source core. CopilotKit also announced a $27M Series A covered by TechCrunch in May 2026, and CEO Atai Barkai published a hands-on short course on DeepLearning.AI covering generative UI with CopilotKit and AG-UI.