AI Chat Client

Conversations that branch like thoughts

A local-first, DAG-based AI chat client. Explore conversations organically, never lose a train of thought.

Get Started

Case — A conversation that branches

User

Design a node-based AI chat flow.

gpt-5.2

Start with data model + interaction rules.

User

03A

Branch: How should context be pinned?

User

03B

Branch: How should nodes be navigated?

gpt-5.2

04A

Context Box + summaries + citations.

glm-4.7

04B

Keyboard-first tree + quick branch compare.

User

Pick a path and keep exploring.

01 — Manifesto

Linear conversations limit thinking. DAGs liberate it.

Traditional chat apps force your thoughts into a straight line. Prompt Tree uses a Directed Acyclic Graph structure, letting conversations branch naturally. Every idea can spawn multiple explorations. No backtracking, no losing context—just pure, unfiltered thinking.

Designed for clarity, built for depth

Every feature serves one purpose: giving you complete control over your AI conversations.

Local-first storage

Your conversations stay on your machine. No cloud dependencies, no privacy concerns, complete ownership of your data.

DAG architecture

Visualize conversations as trees. Branch from any message, compare responses, and never lose context.

Multi-model support

Switch between OpenAI, Anthropic, DeepSeek, and more. Different models for different conversations.

Modern stack

Built with Next.js 16, React 19, and TypeScript. Blazing fast, type-safe, and future-proof.

02 — Concept

The tree is the perfect metaphor for conversation.

A conversation starts with a seed—a prompt. It grows branches—questions, explorations, detours. Some branches wither; others blossom into new insights. Prompt Tree captures this organic process, giving you a visual representation of how ideas evolve and connect.

Install Prompt Tree

Get started in seconds. Available as an npm package for quick installation and updates.

                    # Install globally
                    npm install -g @yxp934/prompt-tree
                    
                    # Or run with npx
                    npx @yxp934/prompt-tree
                    
                    # Start the application
                    prompt-tree
                