Compare
Dify is an LLM application platform: visual workflows, chat and agent orchestration, and built-in RAG all live in one place — the fast route from a model to a working app.
Langhuan does no application orchestration. It turns "document → traceable retrieval" into standalone infrastructure, exposed over REST and MCP over HTTP so any LLM app — Dify included — can use it as one shared knowledge foundation.
| Dimension | 琅嬛 Langhuan | Dify |
|---|---|---|
| Position | Knowledge layer (no orchestration) | LLM app platform (chat/workflow/agents) |
| MCP | Native MCP over HTTP | MCP tool integration inside the platform |
| Chinese keyword search | Native zhparser / gse tokenization + RRF | Vector-dependent, weaker FTS |
| Delivery | Single binary + one .db file (zero-dep SQLite option) | Multi-container suite |
| Traceability | chunk → page/line anchor, full chain | Partial |
| Integration | As the knowledge foundation for your app | Closed loop inside the platform |
If you want to visually build an LLM app, Dify is the more direct path. If you want several apps sharing one traceable knowledge capability, Langhuan is that foundation. They don’t conflict: Langhuan owns the knowledge, Dify owns the orchestration.
Dify is an LLM app platform for workflows, chat, and agent orchestration. Langhuan is a knowledge layer that only does "document → traceable retrieval" and serves it over MCP. One orchestrates, the other grounds.
If several of your apps or agents each wired up their own knowledge base, Langhuan can unify knowledge ingestion and retrieval into one traceable foundation that Dify or your own apps consume, instead of rebuilding it every time.
Two paths: REST (/api/v1/*) or MCP (/mcp). A single binary plus standard PostgreSQL — one docker compose gets you started.
Use document processing and retrieval as a reliable, traceable foundation, then plug it into your product over REST or MCP.
View the project on GitHub