ZENGLIANGYI

What is ZENGLIANGYI?

ZENGLIANGYI (ChatCrystal) is a local-first AI personal knowledge management (PKM) app for developers working with Claude Code, Cursor, Codex CLI, Trae, and GitHub Copilot. It runs as a desktop app (Windows), CLI/web tool, or Docker cloud deployment.

How to use ZENGLIANGYI?

Install the desktop app from GitHub Releases, then configure LLM and embedding providers in Settings and click Import. For CLI use, run npm install -g chatcrystal, then crystal serve -d and crystal import. For Docker, deploy with docker compose up -d, set up a token, and import from devices using crystal connect and crystal import.

Key features of ZENGLIANGYI

• Imports AI coding conversations from multiple tools.
• Distills conversations into structured notes with tags.
• Provides semantic search via embeddings.
• Builds a tag-based knowledge graph with co‑occurrence edges.
• Exposes CLI and MCP tools for agent reuse.
• Runs fully locally with configurable providers.

Use cases of ZENGLIANGYI

• Save and organize conversations from Claude Code, Cursor, Codex CLI, and Copilot.
• Search past AI interactions semantically without relying on cloud services.
• Build a reusable memory layer for AI agents through MCP tools.
• Self‑host a shared knowledge base for a team using Docker cloud.

FAQ from ZENGLIANGYI

What makes ZENGLIANGYI different from other AI note‑taking apps?

It is local‑first and designed specifically for AI coding conversations. It imports from common coding agent tools, distills conversations into structured notes, and exposes an MCP server so agents can recall and write back knowledge.

What platforms and models does ZENGLIANGYI support?

The desktop app runs on Windows. The CLI/web version works on any platform with Node.js ≥ 20. It requires separate LLM and embedding providers; popular models like Claude, GPT, and Qwen are supported for summarization, and any compatible embedding API for search.

Does ZENGLIANGYI support MCP servers?

Yes. Run crystal mcp to start an MCP stdio server that agents can use to recall and store memory.

What is the pricing and licensing for ZENGLIANGYI?

ZENGLIANGYI is open‑source under the Apache License 2.0. There is no pricing; you run it yourself.

Are there any known limitations?

The desktop app is currently only packaged for Windows. The application requires Node.js ≥ 20, and you must configure your own LLM and embedding provider (no built‑in models).