bisheng Documentation

BISHENG is an open LLM devops platform for next generation Enterprise AI applications. Powerful and comprehensive features include: GenAI workflow, RAG,...

11,833 stars TypeScript TypeScript 1 file · ~1,590 tokens #agent#ai#chatbot#enterprise#finetune#genai#gpt#langchian
RAW Doc

Docker & Self-Hosted Deployment Verified OCI

Deploy bisheng on any server, homelab, or cloud container platform.

1-Click Terminal Command
bash (docker run) 
docker run -d --name bisheng -p 3306:3306 --restart always dataelement/bisheng-backend:v3.0.0-beta5
Configured Services & Ports
⚙️ mysql
image: mysql:8.0
🔌 Ports: 3306:3306
⚙️ openfga-migrate
image: cr.dataelem.com/dataelement/openfga:v1.15.2-dm
⚙️ openfga
image: cr.dataelem.com/dataelement/openfga:v1.15.2-dm
🔌 Ports: 8080:8080, 8081:8081, 2112:2112
⚙️ redis
image: redis:7.0.4
🔌 Ports: 6379:6379
⚙️ backend
image: dataelement/bisheng-backend:v3.0.0-beta5
🔌 Ports: 7860:7860
⚙️ backend_worker
image: dataelement/bisheng-backend:v3.0.0-beta5
⚙️ frontend
image: dataelement/bisheng-frontend:v3.0.0-beta5
🔌 Ports: 3001:3001
⚙️ elasticsearch
image: docker.io/bitnamilegacy/elasticsearch:8.12.0
🔌 Ports: 9200:9200, 9300:9300
⚙️ etcd
image: quay.io/coreos/etcd:v3.5.5
⚙️ minio
image: cr.dataelem.com/minio/minio:RELEASE.2023-03-20T20-16-18Z
🔌 Ports: 9100:9000, 9101:9001
⚙️ milvus
image: milvusdb/milvus:v2.5.10
🔌 Ports: 19530:19530, 9091:9091
docker-compose.yml
yaml (docker-compose.yml) 
services:
  mysql:
    container_name: bisheng-mysql
    image: mysql:8.0
    
    ports:
      - "3306:3306"
    environment:
      MYSQL_ROOT_PASSWORD: "1234"  # 数据库密码，如果修改需要同步修改bisheng/congfig/config.yaml配置database_url的mysql连接密码
      MYSQL_DATABASE: bisheng
      TZ: Asia/Shanghai
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/mysql/conf/my.cnf:/etc/mysql/my.cnf
      - ${DOCKER_VOLUME_DIRECTORY:-.}/mysql/data:/var/lib/mysql
    healthcheck:
      test: ["CMD-SHELL", "mysql -u root -p$$MYSQL_ROOT_PASSWORD -e 'CREATE DATABASE IF NOT EXISTS openfga;'"]
      start_period: 30s
      interval: 20s
      timeout: 10s
      retries: 4
    restart: on-failure

  openfga-migrate:
    container_name: bisheng-openfga-migrate
    image: cr.dataelem.com/dataelement/openfga:v1.15.2-dm
    command: migrate
    environment:
      OPENFGA_DATASTORE_ENGINE: mysql
      OPENFGA_DATASTORE_URI: "root:1234@tcp(mysql:3306)/openfga?parseTime=true"
    depends_on:
      mysql:
        condition: service_healthy

  openfga:
    container_name: bisheng-openfga
    image: cr.dataelem.com/dataelement/openfga:v1.15.2-dm
    command: run
    environment:
      OPENFGA_DATASTORE_METRICS_ENABLED: "true"
      OPENFGA_DATASTORE_ENGINE: mysql
      OPENFGA_DATASTORE_URI: "root:1234@tcp(mysql:3306)/openfga?parseTime=true"
      OPENFGA_LOG_FORMAT: json
      OPENFGA_PLAYGROUND_ENABLED: "true"
      OPENFGA_CHECK_QUERY_CACHE_ENABLED: "true"
      OPENFGA_CHECK_QUERY_CACHE_TTL: 30s
      OPENFGA_CHECK_ITERATOR_CACHE_ENABLED: "true"
      OPENFGA_CHECK_ITERATOR_CACHE_TTL: 30s
      OPENFGA_CHECK_ITERATOR_CACHE_MAX_RESULTS: 10000
      OPENFGA_CACHE_CONTROLLER_ENABLED: "true"
      OPENFGA_CACHE_CONTROLLER_TTL: 10s
      OPENFGA_DATASTORE_MAX_OPEN_CONNS: 80
      OPENFGA_DATASTORE_MAX_IDLE_CONNS: 40
      OPENFGA_METRICS_ENABLE_RPC_HISTOGRAMS: "true"
      OPENFGA_RESOLVE_NODE_LIMIT: "${OPENFGA_RESOLVE_NODE_LIMIT:-25}"
      OPENFGA_LIST_OBJECTS_MAX_RESULTS: 1000
    ports:
      - "8080:8080"
      - "8081:8081"
      - "2112:2112"
    depends_on:
      openfga-migrate:
        condition: service_completed_successfully
    # 不在 compose 里写 HEALTHCHECK：openfga 官方镜像为 distroless，无 /bin/sh、无 wget，
    # CMD-SHELL 会永久 unhealthy。未定义时 Compose 将 ``service_healthy`` 视为已启动即可。
    restart: unless-stopped

  redis:
    container_name: bisheng-redis
    image: redis:7.0.4
    ports:
      - "6379:6379"
    environment:
      TZ: Asia/Shanghai
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/data/redis:/data
      - ${DOCKER_VOLUME_DIRECTORY:-.}/redis/redis.conf:/etc/redis.conf
    command: redis-server /etc/redis.conf
    healthcheck:
      test: ["CMD-SHELL", 'redis-cli ping|grep -e "PONG\|NOAUTH"']
      interval: 10s
      timeout: 5s
      retries: 3
    restart: on-failure

  backend:
    container_name: bisheng-backend
    image: dataelement/bisheng-backend:v3.0.0-beta5
    ports:
      - "7860:7860"
    environment:
      TZ: Asia/Shanghai
      BS_SSO_SYNC__GATEWAY_HMAC_SECRET: "bisheng-local-hmac-20260422"
      BS_MILVUS_CONNECTION_ARGS: '{"host":"milvus","port":"19530","user":"","password":"","secure":false}'
      BS_MILVUS_IS_PARTITION: 'true'
      BS_MILVUS_PARTITION_SUFFIX: '1'
      BS_ELASTICSEARCH_URL: 'http://elasticsearch:9200'
      BS_ELASTICSEARCH_SSL_VERIFY: '{}'  # 可根据自己部署的密码进行配置 '{"basic_auth": ("elastic", "elastic")}'
      BS_MINIO_SCHEMA: 'false'
      BS_MINIO_CERT_CHECK: 'false'
      BS_MINIO_ENDPOINT: 'minio:9000'
      BS_MINIO_SHAREPOINT: 'minio:9000'
      BS_MINIO_ACCESS_KEY: 'minioadmin'
      BS_MINIO_SECRET_KEY: 'minioadmin'
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/bisheng/config/config.yaml:/app/bisheng/config.yaml
      - ${DOCKER_VOLUME_DIRECTORY:-.}/bisheng/entrypoint.sh:/app/entrypoint.sh
      - ${DOCKER_VOLUME_DIRECTORY:-.}/data/bisheng:/app/data
    security_opt:
      - seccomp:unconfined
    command: sh entrypoint.sh api  # 启动api服务
    restart: on-failure
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:7860/health"]
      start_period: 30s
      interval: 90s
      timeout: 30s
      retries: 3
    depends_on:
      mysql:
        condition: service_healthy
      redis:
        condition: service_healthy
      # openfga 官方镜像无 HEALTHCHECK，Compose v5 无法用 service_healthy
      openfga:
        condition: service_started

  backend_worker:
    container_name: bisheng-backend-worker
    image: dataelement/bisheng-backend:v3.0.0-beta5
    environment:
      TZ: Asia/Shanghai
      BS_SSO_SYNC__GATEWAY_HMAC_SECRET: "bisheng-local-hmac-20260422"
      BS_MILVUS_CONNECTION_ARGS: '{"host":"milvus","port":"19530","user":"","password":"","secure":false}'
      BS_MILVUS_IS_PARTITION: 'true'
      BS_MILVUS_PARTITION_SUFFIX: '1'
      BS_ELASTICSEARCH_URL: 'http://elasticsearch:9200'
      BS_ELASTICSEARCH_SSL_VERIFY: '{}'  # 可根据自己部署的密码进行配置 '{"basic_auth": ("elastic", "elastic")}'
      BS_MINIO_SCHEMA: 'false'
      BS_MINIO_CERT_CHECK: 'false'
      BS_MINIO_ENDPOINT: 'minio:9000'
      BS_MINIO_SHAREPOINT: 'minio:9000'
      BS_MINIO_ACCESS_KEY: 'minioadmin'
      BS_MINIO_SECRET_KEY: 'minioadmin'
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/bisheng/config/config.yaml:/app/bisheng/config.yaml
      - ${DOCKER_VOLUME_DIRECTORY:-.}/bisheng/entrypoint.sh:/app/entrypoint.sh
      - ${DOCKER_VOLUME_DIRECTORY:-.}/data/bisheng:/app/data
    security_opt:
      - seccomp:unconfined
    command: sh entrypoint.sh worker  # 启动celery的异步worker服务，用来处理一些耗时的任务
    restart: on-failure
    depends_on:
      mysql:
        condition: service_healthy
      redis:
        condition: service_healthy
      openfga:
        condition: service_started

  frontend:
    container_name: bisheng-frontend
    image: dataelement/bisheng-frontend:v3.0.0-beta5
    ports:
      - "3001:3001"
    environment:
      TZ: Asia/Shanghai
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/nginx/nginx.conf:/etc/nginx/nginx.conf
      - ${DOCKER_VOLUME_DIRECTORY:-.}/nginx/conf.d:/etc/nginx/conf.d
    restart: on-failure
    depends_on:
      - backend

  elasticsearch:
    container_name: bisheng-es
    image: docker.io/bitnamilegacy/elasticsearch:8.12.0
    user: root
    ports:
      - "9200:9200"
      - "9300:9300"
    environment:
      TZ: Asia/Shanghai
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/data/es:/bitnami/elasticsearch/data
    restart: on-failure

  etcd:
    container_name: bisheng-milvus-etcd
    image: quay.io/coreos/etcd:v3.5.5
    environment:
      ETCD_AUTO_COMPACTION_MODE: revision
      ETCD_AUTO_COMPACTION_RETENTION: "1000"
      ETCD_QUOTA_BACKEND_BYTES: "4294967296"
      ETCD_SNAPSHOT_COUNT: "50000"
      TZ: Asia/Shanghai
    volumes:
      - ${DOCKER_VOLUME_DIRECTORY:-.}/data/milvus-etcd:/etcd
    command: etcd -advertise-client-urls=http://127.0.0.1:2379 -listen-client-urls http://0.0.0.0:2379 --data-dir /etcd
    restart: on-failure
    healthcheck:
      test: ["CMD", "etcdctl", "endpoint", "health"]
      interval: 30s
      timeout: 20s
      retries: 3

  minio:
    container_name: bisheng-milvus-minio
    image: cr.dataelem.com/minio/minio:RELEASE.2023-03-20T20-16-18Z
    environment:
      MINIO_ACCESS_KEY: minioadmin
      MINIO_SECRET_KEY: minioadmin
    ports:
      - "9100:9000"
      - "9101:9001"
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - ${DOCKER_VOLUME_DIRECTORY:-.}/data/milvus-minio:/minio_data
    command: minio server /minio_data --console-address ":9001"
    restart: on-failure
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9000/minio/health/live"]
      interval: 30s
      timeout: 20s
      retries: 3

  milvus:
    container_name: bisheng-milvus-standalone
    image: milvusdb/milvus:v2.5.10
    command: ["milvus", "run", "standalone"]
    security_opt:
    - seccomp:unconfined
    environment:
      ETCD_ENDPOINTS: etcd:2379
      MINIO_ADDRESS: minio:9000
    volumes:
      - /etc/localtime:/etc/localtime:ro
      - ${DOCKER_VOLUME_DIRECTORY:-.}/data/milvus:/var/lib/milvus
    restart: on-failure
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9091/healthz"]
      start_period: 90s
      interval: 30s
      timeout: 20s
      retries: 3
    ports:
      - "19530:19530"
      - "9091:9091"
    depends_on:
      - etcd
      - minio

File: README.md

Proudly made by Chinese，May we, like the creators of Deepseek and Black Myth: Wukong, bring more wonder and greatness to the world.

源自中国匠心，希望我们能像 [Deepseek]、[黑神话：悟空] 团队一样，给世界带来更多美好。

BISHENG is an open LLM application devops platform, focusing on enterprise scenarios. It has been used by a large number of industry leading organizations and Fortune 500 companies.

"Bi Sheng" was the inventor of movable type printing, which played a vital role in promoting the transmission of human knowledge. We hope that BISHENG can also provide strong support for the widespread implementation of intelligent applications. Everyone is welcome to participate.

Features

  1. Lingsight, a general-purpose agent with expert-level taste: Through the AGL(Agent Guidance Language) framework, we embed domain experts’ preferences, experience, and business logic into the AI, enabling the agent to exhibit “expert-level understanding” when handling tasks.

  2. Unique BISHENG Workflow

    • 🧩 Independent and comprehensive application orchestration framework: Enables the execution of various tasks within a single framework (while similar products rely on bot invocation or separate chatflow and workflow modules for different tasks).
    • 🔄 Human in the loop: Allows users to intervene and provide feedback during the execution of workflows (including multi-turn conversations), whereas similar products can only execute workflows from start to finish without intervention.
    • 💥 Powerful: Supports loops, parallelism, batch processing, conditional logic, and free combination of all logic components. It also handles complex scenarios such as multi-type input/output, report generation, content review, and more.
    • 🖐️ User-friendly and intuitive: Operations like loops, parallelism, and batch processing, which require specialized components in similar products, can be easily visualized in BISHENG as a "flowchart" (drawing a loop forms a loop, aligning elements creates parallelism, and selecting multiple items enables batch processing).

  3. Designed for Enterprise Applications: Document review, fixed-layout report generation, multi-agent collaboration, policy update comparison, support ticket assistance, customer service assistance, meeting minutes generation, resume screening, call record analysis, unstructured data governance, knowledge mining, data analysis, and more.
    The platform supports the construction of highly complex enterprise application scenarios and offers deep optimization with hundreds of components and thousands of parameters.

  4. Enterprise-grade features are the fundamental guarantee for application implementation: security review, RBAC, user group management, traffic control by group, SSO/LDAP, vulnerability scanning and patching, high availability deployment solutions, monitoring, statistics, and more.

  5. High-Precision Document Parsing: Our high-precision document parsing model is trained on a vast amount of high-quality data accumulated over past 5 years. It includes high-precision printed text, handwritten text, and rare character recognition models, table recognition models, layout analysis models, and seal models., table recognition models, layout analysis models, and seal models. You can deploy it privately for free.

  6. A community for sharing best practices across various enterprise scenarios: An open repository of application cases and best practices.

Quick start

Please ensure the following conditions are met before installing BISHENG:

  • CPU >= 4 Virtual Cores
  • RAM >= 16 GB
  • Docker 19.03.9+
  • Docker Compose 1.25.1+

Recommended hardware condition: 18 virtual cores, 48G. In addition to installing BISHENG, we will also install the following third-party components by default: ES, Milvus, and Onlyoffice.

Download BISHENG

bash
git clone https://github.com/dataelement/bisheng.git
# Enter the installation directory
cd bisheng/docker

# If the system does not have the git command, you can download the BISHENG code as a zip file.
wget https://github.com/dataelement/bisheng/archive/refs/heads/main.zip
# Unzip and enter the installation directory
unzip main.zip && cd bisheng-main/docker

Start BISHENG

bash
docker compose -f docker-compose.yml -p bisheng up -d

After the startup is complete, access http://IP:3001 in the browser. The login page will appear, proceed with user registration.

By default, the first registered user will become the system admin.

For more installation and deployment issues, refer to:：Self-hosting

Acknowledgement

This repo benefits from langchain langflow unstructured and LLaMA-Factory . Thanks for their wonderful works.

Thank you to our contributors：

Community & contact

Welcome to join our discussion group

File: docs/README.md

BiSheng 文档导航

找文档从这里开始。原则:规范看根目录、架构看 architecture/、功能文档看 ../features/。

规范(开发前必读)

文档 内容
constitution.md 架构铁律 C1–C7(双 DB / 多租户 / 权限 / 分层 / 错误码 / 安全)——不可违反
SDD-Guide.md 开发流程总纲:流程分级、★ 暂停点、偏差处理、测试分层、harness 现状
../AGENTS.md 全局 agent 规则入口(各子项目规则按目录自动加载)

架构(子系统深度)

architecture/ — 13 篇:总览 / 后端模块 / 工作流引擎 / RAG / Linsight / 双前端 / 数据模型 / 部署 / 开发指南 / 权限 ReBAC / 商业网关 / 多租户 / 游标分页 + 数据库表结构。子系统架构的唯一正文。

功能文档(SDD 产出)

../features/ — 按 v{X.Y.Z}/{NNN}-{name}/ 组织,每个功能含 spec.md / design.md / tasks.md(及按需 testcases.md,如 F025)。模板见 ../features/_templates/。

接口 · 测试 · 部署 · 科普

目录 内容
api/ 接口文档(filelib 纯检索、知识空间/知识库接口)
PRD/ 现行迭代的产品 PRD 与技术方案(按 {版本} {主题} PRD/ 组织)
observability/ BS_METRIC 指标日志契约(监控团队解析依据)
私有化部署/ 部署文档
blog/ 设计科普(如"企业权限体系的前世今生")

Explore Related Technologies & Topics

View All 9,111+ Repositories →

Discover similar high-velocity repositories, agent skills, and OpenAPI specifications across the ecosystem.

#agent Browse agent catalog → #ai Browse ai catalog → #chatbot Browse chatbot catalog → #enterprise Browse enterprise catalog → #finetune Browse finetune catalog → #genai Browse genai catalog → ⚡ MCP Servers Model Context Protocol → 🧠 Agent Skills Prompt & rule files → 🔌 OpenAPI Specs REST & Swagger schemas →