Learn Dify by building a sweet-talker bot and an interview oracle

What it does

self-dify is a Chinese-language tutorial series for building AI applications with Dify. It walks through local deployment via Docker and eleven chapter-based projects that scale from basic prompt design to workflow orchestration, knowledge bases, text-to-SQL, DeepResearch, and MCP. The intended path takes learners from simple interactive agents—like a relationship “sweet-talker” simulator and a new-student guide—up to complex data and research tasks, without demanding deep programming background.

The interesting bit

The curriculum smuggles serious engineering lessons inside quirky consumer apps. A “guess the medical case” game and a Xiaohongshu book-card generator are practical exercises for learning stateful prompts and RAG retrieval, while later chapters treat Dify as a full orchestration layer for databases and multimodal voice agents. It is published by Datawhale, a Chinese open-source learning collective, which keeps the tone pedagogical rather than promotional.

Key highlights

• Eleven star-rated chapters from Docker setup to voice-text agents
• Projects include a relationship simulator, medical case quiz, interview knowledge base, and batch data processing
• Covers Agent workflows, RAG, text-to-SQL with ECharts, DeepResearch, and MCP integration
• Requires only Docker and an LLM API key (or local Ollama); no advanced coding required
• Licensed CC BY-SA 4.0 by the Datawhale community

Caveats

• Entirely Chinese-language; non-readers will need translation support
• This is a curriculum and documentation repo, not a standalone framework or Dify fork
• Advanced chapters assume completion of earlier sections

Verdict

Ideal for Chinese-speaking beginners and intermediates who want a project-driven path into Dify-based LLM development. Pass if you need a standalone tool or English-language documentation.