AI智能体研发之路-工程篇（二）：Dify智能体开发平台一键部署

博客导读：

《AI—工程篇》

AI智能体研发之路-工程篇（一）：Docker助力AI智能体开发提效​​​​​​​

AI智能体研发之路-工程篇（二）：Dify智能体开发平台一键部署

AI智能体研发之路-工程篇（三）：大模型推理服务框架Ollama一键部署

AI智能体研发之路-工程篇（四）：大模型推理服务框架Xinference一键部署

AI智能体研发之路-工程篇（五）：大模型推理服务框架LocalAI一键部署

《AI—模型篇》

AI智能体研发之路-模型篇（一）：大模型训练框架LLaMA-Factory在国内网络环境下的安装、部署及使用

AI智能体研发之路-模型篇（二）：DeepSeek-V2-Chat 训练与推理实战

目录

1.引言

2.docker compose一行命令部署Dify

3.Dify主要特色 

3.1多种大语言模型集成

3.2丰富工具内置+自定义工具支持​

3.3工作流​

3.4Agent 编排​

 4.总结

1.引言

刚刚开始写CSDN，下班了研究研究CSDN的玩法，看到活动区发起了“Agent AI智能体的未来”话题，正好最近工作中比较多的经历集中在这里，所以借着这个话题活动写一些自己的看法。

目前的工作总结来看就是进行AI智能体的开发，组里基于langchain做了一些储备，搭建了内部的一个AI智能体开发平台，支持了知识库、工具调用、多种大模型的配置，但当后来字节的coze、百度的千帆发布，发现AI智能体开发平台真是发展的太快了，相比于coze，丰富的插件工具，稳定的模型，成熟的工作流，分析自己组内的人力情况，根本追不上：组内同学既要全力完成okr，又要每个人开发自己的AI智能体，剩下的精力才能继续完善AI智能体平台。

五一之前，领导在群里扔了一个创业公司基于Dify开发AI应用的介绍。我抱着学习的态度了解了一下Dify：可以称之为开源版的coze！支持多种大模型、插件工具的一键适配，支持知识库自动切割创建索引，支持工具流开发！赞啊，这些功能不正是组内AI智能体开发平台需要添加的功能么。于是马上开始了部署与调试。

先分享dify项目的github：https://github.com/langgenius/dify

“项目由前腾讯云DevOps团队成员创建。我们发现基于OpenAI的API开发GPT应用程序有些麻烦。凭借我们多年在开发者效率工具方面的研发经验，我们希望能够让更多的人使用自然语言开发有趣的应用程序。”

dify由10+全职团队和100+社区贡献者共同维护，迭代非常快，我下载的时候还是0.6.3，现在已经升级为0.6.6，基本上每周都会升级一个版本。这里唠叨两句，对于迭代非常频繁的开源项目，在公司内部使用的时候，咱就先别二次开发了，否则你这边刚开发完一个点，另一边社区升级了，逻辑冲突合不到一块去，尴尬不尴尬。之前听说有个团队基于tensorflow1.x“自研”了深度学习训练框架，多年过去，社区都2.x了，那个团队还在tensorflow1.x上疲于奔命。哈哈哈，扯远了。

说回到本文，标题叫《Agent AI智能体的未来—Dify项目介绍》，其实是有些夸张的，但我认为，了解Dify以及其中依赖的上下游技术、架构，利用Dify快速建立AI智能体demo原型，其实对推进AI智能体开发是有意义的。今天我先简要介绍一下Dify的部署过程，后面的篇幅会具体分享基于Dify的AI智能体开发经验，以及AI智能体开发所依赖的底层技术。

2.docker compose一行命令部署Dify

首先将dify项目下载至服务器上，

git clone https://github.com/langgenius/dify.git

项目主要分为api（后端）和web（前端）两大部分，具体代码后面再分析，进入docker目录

cd docker

目录中包含几个文件，docker-compose.yaml可以通过docker compose直接启动所有服务和依赖，docker-compose.middleware.yaml可以先启动依赖的关系数据库、向量数据库等组件，再单独启动api和web端，隔离做得可以说太棒了。

先看一下docker-compose.yaml的源码

  1. version: '3'
  2. services:
  3.   # API service
  4.   api:
  5.     image: langgenius/dify-api:0.6.6
  6.     restart: always
  7.     environment:
  8.       # Startup mode, 'api' starts the API server.
  9.       MODE: api
  10.       # The log level for the application. Supported values are `DEBUG`, `INFO`, `WARNING`, `ERROR`, `CRITICAL`
  11.       LOG_LEVEL: INFO
  12.       # A secret key that is used for securely signing the session cookie and encrypting sensitive information on the database. You can generate a strong key using `openssl rand -base64 42`.
  13.       SECRET_KEY: sk-9f73s3ljTXVcMT3Blb3ljTqtsKiGHXVcMT3BlbkFJLK7U
  14.       # The base URL of console application web frontend, refers to the Console base URL of WEB service if console domain is
  15.       # different from api or web app domain.
  16.       # example: http://cloud.dify.ai
  17.       CONSOLE_WEB_URL: ''
  18.       # Password for admin user initialization.
  19.       # If left unset, admin user will not be prompted for a password when creating the initial admin account.
  20.       INIT_PASSWORD: ''
  21.       # The base URL of console application api server, refers to the Console base URL of WEB service if console domain is
  22.       # different from api or web app domain.
  23.       # example: http://cloud.dify.ai
  24.       CONSOLE_API_URL: ''
  25.       # The URL prefix for Service API endpoints, refers to the base URL of the current API service if api domain is
  26.       # different from console domain.
  27.       # example: http://api.dify.ai
  28.       SERVICE_API_URL: ''
  29.       # The URL prefix for Web APP frontend, refers to the Web App base URL of WEB service if web app domain is different from
  30.       # console or api domain.
  31.       # example: http://udify.app
  32.       APP_WEB_URL: ''
  33.       # File preview or download Url prefix.
  34.       # used to display File preview or download Url to the front-end or as Multi-model inputs;
  35.       # Url is signed and has expiration time.
  36.       FILES_URL: ''
  37.       # When enabled, migrations will be executed prior to application startup and the application will start after the migrations have completed.
  38.       MIGRATION_ENABLED: 'true'
  39.       # The configurations of postgres database connection.
  40.       # It is consistent with the configuration in the 'db' service below.
  41.       DB_USERNAME: postgres
  42.       DB_PASSWORD: difyai123456
  43.       DB_HOST: db
  44.       DB_PORT: 5432
  45.       DB_DATABASE: dify
  46.       # The configurations of redis connection.
  47.       # It is consistent with the configuration in the 'redis' service below.
  48.       REDIS_HOST: redis
  49.       REDIS_PORT: 6379
  50.       REDIS_USERNAME: ''
  51.       REDIS_PASSWORD: difyai123456
  52.       REDIS_USE_SSL: 'false'
  53.       # use redis db 0 for redis cache
  54.       REDIS_DB: 0
  55.       # The configurations of celery broker.
  56.       # Use redis as the broker, and redis db 1 for celery broker.
  57.       CELERY_BROKER_URL: redis://:difyai123456@redis:6379/1
  58.       # Specifies the allowed origins for cross-origin requests to the Web API, e.g. https://dify.app or * for all origins.
  59.       WEB_API_CORS_ALLOW_ORIGINS: '*'
  60.       # Specifies the allowed origins for cross-origin requests to the console API, e.g. https://cloud.dify.ai or * for all origins.
  61.       CONSOLE_CORS_ALLOW_ORIGINS: '*'
  62.       # CSRF Cookie settings
  63.       # Controls whether a cookie is sent with cross-site requests,
  64.       # providing some protection against cross-site request forgery attacks
  65.       #
  66.       # Default: `SameSite=Lax, Secure=false, HttpOnly=true`
  67.       # This default configuration supports same-origin requests using either HTTP or HTTPS,
  68.       # but does not support cross-origin requests. It is suitable for local debugging purposes.
  69.       #
  70.       # If you want to enable cross-origin support,
  71.       # you must use the HTTPS protocol and set the configuration to `SameSite=None, Secure=true, HttpOnly=true`.
  72.       #
  73.       # The type of storage to use for storing user files. Supported values are `local` and `s3` and `azure-blob` and `google-storage`, Default: `local`
  74.       STORAGE_TYPE: local
  75.       # The path to the local storage directory, the directory relative the root path of API service codes or absolute path. Default: `storage` or `/home/john/storage`.
  76.       # only available when STORAGE_TYPE is `local`.
  77.       STORAGE_LOCAL_PATH: storage
  78.       # The S3 storage configurations, only available when STORAGE_TYPE is `s3`.
  79.       S3_ENDPOINT: 'https://xxx.r2.cloudflarestorage.com'
  80.       S3_BUCKET_NAME: 'difyai'
  81.       S3_ACCESS_KEY: 'ak-difyai'
  82.       S3_SECRET_KEY: 'sk-difyai'
  83.       S3_REGION: 'us-east-1'
  84.       # The Azure Blob storage configurations, only available when STORAGE_TYPE is `azure-blob`.
  85.       AZURE_BLOB_ACCOUNT_NAME: 'difyai'
  86.       AZURE_BLOB_ACCOUNT_KEY: 'difyai'
  87.       AZURE_BLOB_CONTAINER_NAME: 'difyai-container'
  88.       AZURE_BLOB_ACCOUNT_URL: 'https://<your_account_name>.blob.core.windows.net'
  89.       # The Google storage configurations, only available when STORAGE_TYPE is `google-storage`.
  90.       GOOGLE_STORAGE_BUCKET_NAME: 'yout-bucket-name'
  91.       GOOGLE_STORAGE_SERVICE_ACCOUNT_JSON_BASE64: 'your-google-service-account-json-base64-string'
  92.       # The type of vector store to use. Supported values are `weaviate`, `qdrant`, `milvus`, `relyt`.
  93.       VECTOR_STORE: weaviate
  94.       # The Weaviate endpoint URL. Only available when VECTOR_STORE is `weaviate`.
  95.       WEAVIATE_ENDPOINT: http://weaviate:8080
  96.       # The Weaviate API key.
  97.       WEAVIATE_API_KEY: WVF5YThaHlkYwhGUSmCRgsX3tD5ngdN8pkih
  98.       # The Qdrant endpoint URL. Only available when VECTOR_STORE is `qdrant`.
  99.       QDRANT_URL: http://qdrant:6333
  100.       # The Qdrant API key.
  101.       QDRANT_API_KEY: difyai123456
  102.       # The Qdrant client timeout setting.
  103.       QDRANT_CLIENT_TIMEOUT: 20
  104.       # The Qdrant client enable gRPC mode.
  105.       QDRANT_GRPC_ENABLED: 'false'
  106.       # The Qdrant server gRPC mode PORT.
  107.       QDRANT_GRPC_PORT: 6334
  108.       # Milvus configuration Only available when VECTOR_STORE is `milvus`.
  109.       # The milvus host.
  110.       MILVUS_HOST: 127.0.0.1
  111.       # The milvus host.
  112.       MILVUS_PORT: 19530
  113.       # The milvus username.
  114.       MILVUS_USER: root
  115.       # The milvus password.
  116.       MILVUS_PASSWORD: Milvus
  117.       # The milvus tls switch.
  118.       MILVUS_SECURE: 'false'
  119.       # relyt configurations
  120.       RELYT_HOST: db
  121.       RELYT_PORT: 5432
  122.       RELYT_USER: postgres
  123.       RELYT_PASSWORD: difyai123456
  124.       RELYT_DATABASE: postgres
  125.       # Mail configuration, support: resend, smtp
  126.       MAIL_TYPE: ''
  127.       # default send from email address, if not specified
  128.       MAIL_DEFAULT_SEND_FROM: 'YOUR EMAIL FROM (eg: no-reply )'
  129.       SMTP_SERVER: ''
  130.       SMTP_PORT: 587
  131.       SMTP_USERNAME: ''
  132.       SMTP_PASSWORD: ''
  133.       SMTP_USE_TLS: 'true'
  134.       # the api-key for resend (https://resend.com)
  135.       RESEND_API_KEY: ''
  136.       RESEND_API_URL: https://api.resend.com
  137.       # The DSN for Sentry error reporting. If not set, Sentry error reporting will be disabled.
  138.       SENTRY_DSN: ''
  139.       # The sample rate for Sentry events. Default: `1.0`
  140.       SENTRY_TRACES_SAMPLE_RATE: 1.0
  141.       # The sample rate for Sentry profiles. Default: `1.0`
  142.       SENTRY_PROFILES_SAMPLE_RATE: 1.0
  143.       # Notion import configuration, support public and internal
  144.       NOTION_INTEGRATION_TYPE: public
  145.       NOTION_CLIENT_SECRET: you-client-secret
  146.       NOTION_CLIENT_ID: you-client-id
  147.       NOTION_INTERNAL_SECRET: you-internal-secret
  148.       # The sandbox service endpoint.
  149.       CODE_EXECUTION_ENDPOINT: "http://sandbox:8194"
  150.       CODE_EXECUTION_API_KEY: dify-sandbox
  151.       CODE_MAX_NUMBER: 9223372036854775807
  152.       CODE_MIN_NUMBER: -9223372036854775808
  153.       CODE_MAX_STRING_LENGTH: 80000
  154.       TEMPLATE_TRANSFORM_MAX_LENGTH: 80000
  155.       CODE_MAX_STRING_ARRAY_LENGTH: 30
  156.       CODE_MAX_OBJECT_ARRAY_LENGTH: 30
  157.       CODE_MAX_NUMBER_ARRAY_LENGTH: 1000
  158.     depends_on:
  159.       - db
  160.       - redis
  161.     volumes:
  162.       # Mount the storage directory to the container, for storing user files.
  163.       - ./volumes/app/storage:/app/api/storage
  164.     # uncomment to expose dify-api port to host
  165.     # ports:
  166.     #   - "5001:5001"
  167.  
  168.   # worker service
  169.   # The Celery worker for processing the queue.
  170.   worker:
  171.     image: langgenius/dify-api:0.6.6
  172.     restart: always
  173.     environment:
  174.       # Startup mode, 'worker' starts the Celery worker for processing the queue.
  175.       MODE: worker
  176.  
  177.       # --- All the configurations below are the same as those in the 'api' service. ---
  178.  
  179.       # The log level for the application. Supported values are `DEBUG`, `INFO`, `WARNING`, `ERROR`, `CRITICAL`
  180.       LOG_LEVEL: INFO
  181.       # A secret key that is used for securely signing the session cookie and encrypting sensitive information on the database. You can generate a strong key using `openssl rand -base64 42`.
  182.       # same as the API service
  183.       SECRET_KEY: sk-9f73s3ljTXVcMT3Blb3ljTqtsKiGHXVcMT3BlbkFJLK7U
  184.       # The configurations of postgres database connection.
  185.       # It is consistent with the configuration in the 'db' service below.
  186.       DB_USERNAME: postgres
  187.       DB_PASSWORD: difyai123456
  188.       DB_HOST: db
  189.       DB_PORT: 5432
  190.       DB_DATABASE: dify
  191.       # The configurations of redis cache connection.
  192.       REDIS_HOST: redis
  193.       REDIS_PORT: 6379
  194.       REDIS_USERNAME: ''
  195.       REDIS_PASSWORD: difyai123456
  196.       REDIS_DB: 0
  197.       REDIS_USE_SSL: 'false'
  198.       # The configurations of celery broker.
  199.       CELERY_BROKER_URL: redis://:difyai123456@redis:6379/1
  200.       # The type of storage to use for storing user files. Supported values are `local` and `s3` and `azure-blob`, Default: `local`
  201.       STORAGE_TYPE: local
  202.       STORAGE_LOCAL_PATH: storage
  203.       # The S3 storage configurations, only available when STORAGE_TYPE is `s3`.
  204.       S3_ENDPOINT: 'https://xxx.r2.cloudflarestorage.com'
  205.       S3_BUCKET_NAME: 'difyai'
  206.       S3_ACCESS_KEY: 'ak-difyai'
  207.       S3_SECRET_KEY: 'sk-difyai'
  208.       S3_REGION: 'us-east-1'
  209.       # The Azure Blob storage configurations, only available when STORAGE_TYPE is `azure-blob`.
  210.       AZURE_BLOB_ACCOUNT_NAME: 'difyai'
  211.       AZURE_BLOB_ACCOUNT_KEY: 'difyai'
  212.       AZURE_BLOB_CONTAINER_NAME: 'difyai-container'
  213.       AZURE_BLOB_ACCOUNT_URL: 'https://<your_account_name>.blob.core.windows.net'
  214.       # The type of vector store to use. Supported values are `weaviate`, `qdrant`, `milvus`, `relyt`.
  215.       VECTOR_STORE: weaviate
  216.       # The Weaviate endpoint URL. Only available when VECTOR_STORE is `weaviate`.
  217.       WEAVIATE_ENDPOINT: http://weaviate:8080
  218.       # The Weaviate API key.
  219.       WEAVIATE_API_KEY: WVF5YThaHlkYwhGUSmCRgsX3tD5ngdN8pkih
  220.       # The Qdrant endpoint URL. Only available when VECTOR_STORE is `qdrant`.
  221.       QDRANT_URL: http://qdrant:6333
  222.       # The Qdrant API key.
  223.       QDRANT_API_KEY: difyai123456
  224.       # The Qdrant clinet timeout setting.
  225.       QDRANT_CLIENT_TIMEOUT: 20
  226.       # The Qdrant client enable gRPC mode.
  227.       QDRANT_GRPC_ENABLED: 'false'
  228.       # The Qdrant server gRPC mode PORT.
  229.       QDRANT_GRPC_PORT: 6334
  230.       # Milvus configuration Only available when VECTOR_STORE is `milvus`.
  231.       # The milvus host.
  232.       MILVUS_HOST: 127.0.0.1
  233.       # The milvus host.
  234.       MILVUS_PORT: 19530
  235.       # The milvus username.
  236.       MILVUS_USER: root
  237.       # The milvus password.
  238.       MILVUS_PASSWORD: Milvus
  239.       # The milvus tls switch.
  240.       MILVUS_SECURE: 'false'
  241.       # Mail configuration, support: resend
  242.       MAIL_TYPE: ''
  243.       # default send from email address, if not specified
  244.       MAIL_DEFAULT_SEND_FROM: 'YOUR EMAIL FROM (eg: no-reply )'
  245.       # the api-key for resend (https://resend.com)
  246.       RESEND_API_KEY: ''
  247.       RESEND_API_URL: https://api.resend.com
  248.       # relyt configurations
  249.       RELYT_HOST: db
  250.       RELYT_PORT: 5432
  251.       RELYT_USER: postgres
  252.       RELYT_PASSWORD: difyai123456
  253.       RELYT_DATABASE: postgres
  254.       # Notion import configuration, support public and internal
  255.       NOTION_INTEGRATION_TYPE: public
  256.       NOTION_CLIENT_SECRET: you-client-secret
  257.       NOTION_CLIENT_ID: you-client-id
  258.       NOTION_INTERNAL_SECRET: you-internal-secret
  259.     depends_on:
  260.       - db
  261.       - redis
  262.     volumes:
  263.       # Mount the storage directory to the container, for storing user files.
  264.       - ./volumes/app/storage:/app/api/storage
  265.  
  266.   # Frontend web application.
  267.   web:
  268.     image: langgenius/dify-web:0.6.6
  269.     restart: always
  270.     environment:
  271.       # The base URL of console application api server, refers to the Console base URL of WEB service if console domain is
  272.       # different from api or web app domain.
  273.       # example: http://cloud.dify.ai
  274.       CONSOLE_API_URL: ''
  275.       # The URL for Web APP api server, refers to the Web App base URL of WEB service if web app domain is different from
  276.       # console or api domain.
  277.       # example: http://udify.app
  278.       APP_API_URL: ''
  279.       # The DSN for Sentry error reporting. If not set, Sentry error reporting will be disabled.
  280.       SENTRY_DSN: ''
  281.     # uncomment to expose dify-web port to host
  282.     # ports:
  283.     #   - "3000:3000"
  284.  
  285.   # The postgres database.
  286.   db:
  287.     image: postgres:15-alpine
  288.     restart: always
  289.     environment:
  290.       PGUSER: postgres
  291.       # The password for the default postgres user.
  292.       POSTGRES_PASSWORD: difyai123456
  293.       # The name of the default postgres database.
  294.       POSTGRES_DB: dify
  295.       # postgres data directory
  296.       PGDATA: /var/lib/postgresql/data/pgdata
  297.     volumes:
  298.       - ./volumes/db/data:/var/lib/postgresql/data
  299.     # uncomment to expose db(postgresql) port to host
  300.     # ports:
  301.     #   - "5432:5432"
  302.     healthcheck:
  303.       test: [ "CMD", "pg_isready" ]
  304.       interval: 1s
  305.       timeout: 3s
  306.       retries: 30
  307.  
  308.   # The redis cache.
  309.   redis:
  310.     image: redis:6-alpine
  311.     restart: always
  312.     volumes:
  313.       # Mount the redis data directory to the container.
  314.       - ./volumes/redis/data:/data
  315.     # Set the redis password when startup redis server.
  316.     command: redis-server --requirepass difyai123456
  317.     healthcheck:
  318.       test: [ "CMD", "redis-cli", "ping" ]
  319.     # uncomment to expose redis port to host
  320.     # ports:
  321.     #   - "6379:6379"
  322.  
  323.   # The Weaviate vector store.
  324.   weaviate:
  325.     image: semitechnologies/weaviate:1.19.0
  326.     restart: always
  327.     volumes:
  328.       # Mount the Weaviate data directory to the container.
  329.       - ./volumes/weaviate:/var/lib/weaviate
  330.     environment:
  331.       # The Weaviate configurations
  332.       # You can refer to the [Weaviate](https://weaviate.io/developers/weaviate/config-refs/env-vars) documentation for more information.
  333.       QUERY_DEFAULTS_LIMIT: 25
  334.       AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED: 'false'
  335.       PERSISTENCE_DATA_PATH: '/var/lib/weaviate'
  336.       DEFAULT_VECTORIZER_MODULE: 'none'
  337.       CLUSTER_HOSTNAME: 'node1'
  338.       AUTHENTICATION_APIKEY_ENABLED: 'true'
  339.       AUTHENTICATION_APIKEY_ALLOWED_KEYS: 'WVF5YThaHlkYwhGUSmCRgsX3tD5ngdN8pkih'
  340.       AUTHENTICATION_APIKEY_USERS: '[email protected]'
  341.       AUTHORIZATION_ADMINLIST_ENABLED: 'true'
  342.       AUTHORIZATION_ADMINLIST_USERS: '[email protected]'
  343.     # uncomment to expose weaviate port to host
  344.     # ports:
  345.     #  - "8080:8080"
  346.  
  347.   # The DifySandbox
  348.   sandbox:
  349.     image: langgenius/dify-sandbox:0.1.0
  350.     restart: always
  351.     cap_add:
  352.     # Why is sys_admin permission needed?
  353.     # https://docs.dify.ai/getting-started/install-self-hosted/install-faq#id-16.-why-is-sys_admin-permission-needed
  354.       - SYS_ADMIN
  355.     environment:
  356.       # The DifySandbox configurations
  357.       API_KEY: dify-sandbox
  358.       GIN_MODE: release
  359.       WORKER_TIMEOUT: 15
  360.  
  361.   # Qdrant vector store.
  362.   # uncomment to use qdrant as vector store.
  363.   # (if uncommented, you need to comment out the weaviate service above,
  364.   # and set VECTOR_STORE to qdrant in the api & worker service.)
  365.   # qdrant:
  366.   #   image: langgenius/qdrant:v1.7.3
  367.   #   restart: always
  368.   #   volumes:
  369.   #     - ./volumes/qdrant:/qdrant/storage
  370.   #   environment:
  371.   #     QDRANT_API_KEY: 'difyai123456'
  372.   #   # uncomment to expose qdrant port to host
  373.   #   # ports:
  374.   #   #  - "6333:6333"
  375.   #   #  - "6334:6334"
  376.  
  377.   # The nginx reverse proxy.
  378.   # used for reverse proxying the API service and Web service.
  379.   nginx:
  380.     image: nginx:latest
  381.     restart: always
  382.     volumes:
  383.       - ./nginx/nginx.conf:/etc/nginx/nginx.conf
  384.       - ./nginx/proxy.conf:/etc/nginx/proxy.conf
  385.       - ./nginx/conf.d:/etc/nginx/conf.d
  386.       #- ./nginx/ssl:/etc/ssl
  387.     depends_on:
  388.       - api
  389.       - web
  390.     ports:
  391.       - "80:80"
  392.       #- "443:443"

主要包含如下几个模块及docker镜像

  • api：langgenius/dify-api:0.6.6
  • worker：langgenius/dify-api:0.6.6
  • web：langgenius/dify-web:0.6.6
  • db：postgres:15-alpine
  • redis：redis:6-alpine
  • weaviate：semitechnologies/weaviate:1.19.0
  • sandbox：langgenius/dify-sandbox:0.1.0
  • nginx：nginx:latest

docker compose一键部署安装

docker compose up -d

从dockerhub逐个pulling依赖镜像，感觉还挺爽的

等待镜像下载部署完成后，打开webui：123.123.123.123:80，默认占用80端口，可以在docker compose配置文件中更改nignx端口。

经过账号初始化等工作后，欢迎来到Dify工作界面，就是这么丝滑

3.Dify主要特色 

3.1多种大语言模型集成

相较于fastGPT+oneApi的方案，集成度更高：

  • 大模型厂商模型：仅需到大模型厂商注册账号，申请鉴权apikey，即可快速体验比较每个大模型厂商的优劣，项目还贴心的附带了每个厂商的注册跳转链接，每个厂商基本都会给几百万的token用于测试。
  • 本地部署模型：支持Xinference，Ollama，OpenLLM，LocalAI等推理框架部署的模型一键接入
  • ​HuggingFace开源模型：只需配置APIKEY和模型名字即可接入，不过要求服务器能翻墙噢
  • OpenAI-API-compatible：接入兼容OpenAI规范的API，目前Xinference，OpenLLM等很多推理框架，都是直接支持OpenAI API接口规范的，但对于每家大模型厂商，一般都要设计自己的规范，可能是想构建生态，或者是与众不同？这里多说一下：对于国内互联网中小厂，如果不自建大模型，可能要试用或买入多加公司的大模型，多加大模型给多个业务部门使用，就要有一个代理平台专门计算成本，这个平台最好对业务暴露的是兼容OpenAI的API接口，如果不是的话，当接入Dify平台时，就需要包一层与OpenAI兼容的API接口。

3.2丰富工具内置+自定义工具支持​

 Dify内置了包含搜索引擎、天气预报、维基百科、SD等工具，同时自定义工具的配置化接入，团队成员一人接入，全组复用，高效！

3.3工作流​

 只需连接各个节点，既能在几分钟内快速完成AI智能体创作，且逻辑非常清晰。

3.4Agent 编排​

编写提示词，导入知识库，添加工具，选择模型，运行测试，发布为API，一条龙创作！ 

还有很多特色，在此不再赘述了，附上一张官方的表格吧

 4.总结

临下班了，本来只是想参加个话题活动，洋洋洒洒几千字。可能加入了太多感慨吧。希望感兴趣的朋友可以关注我、点赞、收藏和评论，您的鼓励是我持续码字的动力。

本文首先结合自己的工作写了一些对Agent AI智能体的见解，接着介绍了Dify框架快捷部署的过程，最后阐述了Dify框架的特点。个人认为Dify的发展会让Agent AI智能体开发提效，涌现更多有趣有价值的AI应用。

最后，写一下我对未来AI智能体发展的看法吧，从流量与用户来看，2000年-2004年，以新浪、搜狐、网易为代表的门户网站是流量入口，2004年-2014年，以百度为代表的搜索引擎是流量入口，2014-2024，以抖音、快手、微博、小红书为代表的移动互联网推荐系统是流量入口，抓住了流量入口就抓住了用户，抓住了用户就抓住了商业变现。2024-未来，极大的可能出现一家基于AI的平台型企业，通过AI智能体抓住流量，比如你要去哪玩，AI智能体在为你做出规划的过程中，夹杂酒店、航班的广告私货，你想吃什么，AI智能体夹杂着饭店的广告私货。

AI领域，乾坤未定，你我皆是黑马。

如果对AI感兴趣，可以接着看看我的其他文章：

《AI—工程篇》

AI智能体研发之路-工程篇（一）：Docker助力AI智能体开发提效​​​​​​​

AI智能体研发之路-工程篇（二）：Dify智能体开发平台一键部署

AI智能体研发之路-工程篇（三）：大模型推理服务框架Ollama一键部署

AI智能体研发之路-工程篇（四）：大模型推理服务框架Xinference一键部署

AI智能体研发之路-工程篇（五）：大模型推理服务框架LocalAI一键部署

《AI—模型篇》

AI智能体研发之路-模型篇（一）：大模型训练框架LLaMA-Factory在国内网络环境下的安装、部署及使用

AI智能体研发之路-模型篇（二）：DeepSeek-V2-Chat 训练与推理实战

文章知识点与官方知识档案匹配，可进一步学习相关知识
Python入门技能树首页概览423497 人正在系统学习中
注：本文转载自blog.csdn.net的LDG_AGI的文章"https://blog.csdn.net/weixin_48007632/article/details/138543709"。版权归原作者所有，此博客不拥有其著作权，亦不承担相应法律责任。如有侵权，请联系我们删除。
复制链接
相关推荐
发表评论

评论记录：

未查询到任何数据！