Kenhuang Academy
  • English

      • Resources

      Dify moves quickly (nodes, Marketplace plugins, Agent shapes). Prefer docs.dify.ai. The links below are what this tutorial was checked against, plus the sibling courses on Kenhuang Academy.

      Official

      ResourceURL
      Docs homehttps://docs.dify.ai
      Docs index (llms.txt)https://docs.dify.ai/llms.txt
      Self-host Docker Composehttps://docs.dify.ai/en/self-host/deploy/quick-start/docker-compose
      Cloudhttps://cloud.dify.ai
      GitHubhttps://github.com/langgenius/dify
      Send Chat Messagehttps://docs.dify.ai/en/api-reference/chat-messages/send-chat-message
      CLI difyctlhttps://docs.dify.ai/en/cli/overview
      Model providershttps://docs.dify.ai/en/self-host/use-dify/workspace/model-providers
      Key concepts (app types)https://docs.dify.ai/en/learn/key-concepts
      Discord / MarketplaceExtra section on the docs home, plus in-product Marketplace

      Chinese docs live on the same site under /zh/ paths. Plugins, tools, and some providers come from the Marketplace—they are not a hard-coded core-repo list.

      Hardware and Compose steps follow the self-host page: CPU ≥ 2, RAM ≥ 4 GiB, Mac VM 2 vCPU / 8 GiB, Compose 2.24.0+.

      Doc paths worth bookmarking

      TopicWhere on the docs site
      App typesLearn → Key Concepts; Use → Chatbot / Agent / Workflow
      KnowledgeUse → Knowledge (chunking, High Quality, hybrid retrieval)
      PublishUse → Publish (Web App, API, MCP Server)
      Self-host envSelf-Host → Configuration → Environment Variables
      Guided lessonsLearn → Tutorials (customer-service bot, Workflow 101)
      TutorialWhen to open it
      OllamaPull local models, then Dify’s Ollama chapter
      LangChainSame RAG / agent ideas as testable Python
      MCPAttach a Dify app or other tools to Cursor-class clients
      DockerCompose, volumes, networks—self-host debugging

      Dify is visual; LangChain is code-first; Ollama serves local weights; MCP is how those capabilities show up in an IDE. Real projects often use all four.

      Tutorial index

      1. Dify Tutorial
      2. Introduction
      3. Installation
      4. Quick Start
      5. Model Providers
      6. Knowledge Base
      7. Workflow
      8. Agent
      9. Publish & API
      10. Ollama Integration
      11. Practical Examples

      Community and versions

      • Bugs and ideas: GitHub Issues / Discussions (langgenius/dify)
      • Self-host upgrades: back up docker/volumes and .env, then diff the new .env.example. Do not freeze an old docker-compose.yaml forever
      • This tutorial follows public Community Edition + Cloud docs. Enterprise screens may differ; trust the docs shipped with your build

      Four lines of architecture: Dify orchestrates and publishes; LangChain holds logic you want under unit tests; Ollama serves local weights; MCP is how the same capability shows up in an IDE.

      Contribute

      This guide lives in Kenhuang Academy. When labels, Compose service names, or API fields change, open an issue or PR and update both docs/zh/dify/ and docs/en/dify/.

      评论