智川X-Agent是中科闻歌推出的一站式企业智能体开发平台，帮助企业零代码快速构建AI应用。智川X-Agent基于封装大模型、知识库、工作流等复杂技术模块为可视化组件，用户通过简单的拖拽和配置可搭建符合业务需求的AI应用。智川X-Agent平台支持多种大模型（如雅意、文心一言等），提供知识库管理、工作流编排、应用发布等功能，满足政务、金融、媒体等多行业需求，助力企业实现AI应用的极速落地与高效迭代，加速AI普惠化。
## 快速开始
通过docker 镜像启动 x-agent:
docker volume create agent-x-data
docker volume create agent-x-cicd
docker pull ccr.ccs.tencentyun.com/wenge/agent-x:agent-x_no_bge_250815_05
docker run -d --restart=always -p 80:80 -p 443:443 -p 8848:8848 -p 3306:3306 -p 6379:6379 -p 9200:9200 -p 9000:9000 -p 9001:9001 -e IP_ADDR="127.0.0.1:80" -v agent-x-data:/u01/isi -v agent-x-cicd:/app/agent/server --name agent-x ccr.ccs.tencentyun.com/wenge/agent-x:agent-x_no_bge_250815_05访问管理后台 http://127.0.0.1:80/wg-agent-manage/#/appmanage
- 账号: agent-x
- 密码: 04p9xa0gAE*%&Op8
- 账号: nacos
- 密码: k2j210w5CKKO!&Wh0
127.0.0.1:3306
- 账号: root
- 密码: 2ievD%GBA6
- 主库：smart_customer_agent
- 账号: admin
- 密码: 6838BHE%%C472
docker pull ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2
mkdir -p /u01/isi/code_sdk
#下载 算法配置文件包并上传到服务器指定目录，配置文件包在目录：/config/Agent_X.zip
unzip Agent_X.zipdocker run -d -v /u01/isi/code_sdk/Embedding_model/config.yml:/app/config.yml -v /u01/isi/code_sdk/Embedding_model/main.py:/app/main.py -p 10822:8080 ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2docker run -d -v /u01/isi/code_sdk/Code_node/main.py:/app/main.py -p 1216:8080 ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2docker run -d -v /u01/isi/code_sdk/Nl2sql/config.yaml:/app/config.yaml -v /u01/isi/code_sdk/Nl2sql/main.py:/app/main.py -p 1025:8080 ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2docker run -d -v /u01/isi/code_sdk/MCP/config.yml:/app/config.yml -v /u01/isi/code_sdk/MCP/main.py:/app/main.py -p 4011:8080 ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2docker run -d --name algorithm-local-mcp --net host -v /u01/isi/code_sdk/Auto_mcp/config.yml:/app/config.yml -v /u01/isi/code_sdk/Auto_mcp/main.py:/app/main.py -v /u01/isi/code_sdk/Auto_mcp/mcp_file:/app/mcp_file ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2docker run -d -v /u01/isi/code_sdk/mcp_sql/config.yaml:/app/config.yaml -v /u01/isi/code_sdk/mcp_sql/main.py:/app/main.py -p 4016:8080 ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2docker pull ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_chrome_v2docker run -d -v /u01/isi/code_sdk/URL_analysis/config.yml:/app/config.yml -v /u01/isi/code_sdk/URL_analysis/main.py:/app/main.py -p 9007:8080 ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_chrome_v2docker run -d -v /u01/isi/code_sdk/URL_to_img/config.yml:/app/config.yml -v /u01/isi/code_sdk/URL_to_img/main.py:/app/main.py -p 5028:8080 ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_chrome_v2docker pull ccr.ccs.tencentyun.com/wenge/agent-x:reranker.v1
docker run -d -p 9098:8080 --restart=always --name reranker ccr.ccs.tencentyun.com/wenge/agent-x:reranker.v1docker pull ccr.ccs.tencentyun.com/wenge/agent-x:contentparse.v4.7
docker run -d -p 9099:8080 --restart=always --name content_parse ccr.ccs.tencentyun.com/wenge/agent-x:contentparse.v4.7docker pull ccr.ccs.tencentyun.com/wenge/agent-x:doc_answer_noes_nosql.v1.2.8-build2503143-encrypted
docker run -d -p 9097:8080 --restart=always --name yayi-plugin-doc-answer-250623 ccr.ccs.tencentyun.com/wenge/agent-x:doc_answer_noes_nosql.v1.2.8-build2503143-encrypted注意：如果默认服务的 ip 和端口有变动，请调整对应的配置项！
appframe:
yayi:
# 10.文档智能解析
contentparsingnewversion:
uri: http://172.17.0.0.1:9099/analysis
# 9.重排序服务
rearrange:
uri: http://172.17.0.0.1:9098/analysis
# 11.文档切片
knowledgesplit:
uri: http://172.17.0.1:9097/analysis
# 网页快照
screenshot:
# 图片上传
uploadUrl: http://172.17.0.1:80/smart-agent-api/wos/file/upload
# 8.网页截图
api: http://172.17.0.1:5028/capture-screenshot
# mcp服务api
mcp:
# 4.MCP
serviceApi: http://172.17.0.1:4011/service
# 4.MCP
queryApi: http://172.17.0.1:4011/query
# 5.本地自定义安装 mcp
buildMcpApi: http://172.17.0.1:4011/deploy_service
# 3.智能问数（NL2SQL）
textToSqlSse: http://172.17.0.1:1025/get_answer_text2sql
workflow:
default:
# 2.工作流代码节点
codeApi: http://172.17.0.1:1216/execute
# 开始节点的改写时使用的大模型 id（llm_info表的 model_id字段）
startRewriteModelId: 87026c3464664ad49a8b622ec719fa70-- 1.启动向量模型
use smart_customer_agent;
update smart_customer_agent.dense_vector set uri='http://172.17.0.1:10822/analysis' where code = 'local_bge_768';- 前端： vue
- 后端： java
- 算法： python
- 数据库： mysql
- 中间件：elasticsearch,redis,minio,nginx
version: '3.8'
services:
# ========== 主服务：x-agent ==========
agent-x:
image: ccr.ccs.tencentyun.com/wenge/agent-x:agent-x_no_bge_250815_05
container_name: agent-x
restart: always
ports:
- "80:80" # 替换为浏览器将要访问的端口，与IP_ADDR参数的端口一致
- "443:443" # https
- "8848:8848" # Nacos
- "3306:3306" # MySQL
- "6379:6379" # Redis
- "9200:9200" # Elasticsearch
- "9000:9000" # MinIO
- "9001:9001" # MinIO
environment:
- IP_ADDR=127.0.0.1:80 # 替换为浏览器将要访问的 ip 和端口
volumes:
- agent-x-data:/u01/isi
- agent-x-cicd:/app/agent/server
# ========== 算法服务 ==========
# 1. 向量模型
algorithm-vector:
image: ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2
container_name: algorithm-vector
restart: always
ports:
- "10822:8080" # 默认端口10822，如果默认端口10822有变动，请修改 mysql 的配置项:use smart_customer_agent; update smart_customer_agent.dense_vector set uri='http://172.17.0.1:10822/analysis' where code = 'local_bge_768';
volumes:
- ./code_sdk/Embedding_model/config.yml:/app/config.yml
- ./code_sdk/Embedding_model/main.py:/app/main.py
# 2. 工作流代码节点
algorithm-code-node:
image: ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2
container_name: algorithm-code-node
restart: always
ports:
- "1216:8080" #默认端口1216，如果默认端口1216有变动，请修改 nacos 的配置项: workflow.default.codeApi: http://172.17.0.1:1216/execute
volumes:
- ./code_sdk/Code_node/main.py:/app/main.py
# 3. 智能问数（NL2SQL）
algorithm-nl2sql:
image: ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2
container_name: algorithm-nl2sql
restart: always
ports:
- "1025:8080" #默认端口1025，如果默认端口1025有变动，请修改 nacos 的配置项: mcp.textToSqlSse: http://172.17.0.1:1025/get_answer_text2sql
volumes:
- ./code_sdk/Nl2sql/config.yaml:/app/config.yaml
- ./code_sdk/Nl2sql/main.py:/app/main.py
# 4. MCP
algorithm-mcp:
image: ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2
container_name: algorithm-mcp
restart: always
ports:
- "4011:8080" #默认端口4011，如果默认端口4011有变动，请修改 nacos 的配置项: mcp.buildMcpApi: http://172.17.0.1:4011/deploy_service
volumes:
- ./code_sdk/MCP/config.yml:/app/config.yml
- ./code_sdk/MCP/main.py:/app/main.py
# 5. 本地自定义 MCP (使用 host 网络)
algorithm-local-mcp:
image: ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_v2
container_name: algorithm-local-mcp
restart: always
network_mode: host # 使用宿主机网络
volumes:
- ./code_sdk/Auto_mcp/config.yml:/app/config.yml
- ./code_sdk/Auto_mcp/main.py:/app/main.py
- ./code_sdk/Auto_mcp/mcp_file:/app/mcp_file
# 8. 网页截图
algorithm-url-to-img:
image: ccr.ccs.tencentyun.com/wenge/agent-x:algorithm_chrome_v2
container_name: algorithm-url-to-img
restart: always
ports:
- "5028:8080" #默认端口5028，如果默认端口5028有变动，请修改 nacos 的配置项: screenshot.api: http://172.17.0.1:5028/capture-screenshot
volumes:
- ./code_sdk/URL_to_img/config.yml:/app/config.yml
- ./code_sdk/URL_to_img/main.py:/app/main.py
# 9. 重排序服务 (yayi)
algorithm-reranker:
image: ccr.ccs.tencentyun.com/wenge/agent-x:reranker.v1
container_name: algorithm-reranker
restart: always
ports:
- "9098:8080" #默认端口9098，如果默认端口9098有变动，请修改 nacos 的配置项: appframe.yayi.rearrange.uri: http://172.17.0.0.1:9098/analysis
# 10. 文档智能解析 (yayi)
algorithm-content-parse:
image: ccr.ccs.tencentyun.com/wenge/agent-x:contentparse.v4.7
container_name: algorithm-content-parse
restart: always
ports:
- "9099:8080" #默认端口9099，如果默认端口9099有变动，请修改 nacos 的配置项: appframe.yayi.contentparsingnewversion.uri: http://172.17.0.0.1:9099/analysis
doc_answer_noes:
image: ccr.ccs.tencentyun.com/wenge/agent-x:doc_answer_noes_nosql.v1.2.8-build2503143-encrypted
container_name: doc_answer_noes
restart: always
ports:
- "9097:8080" #默认端口9097，如果默认端口9097有变动，请修改 nacos 的配置项: appframe.yayi.knowledgesplit.uri: http://172.17.0.0.1:9097/analysis
# ========== 数据卷定义 ==========
volumes:
agent-x-data:
external: true # 对应你之前创建的 docker volume create agent-x-data
agent-x-cicd:
external: true # 对应你之前创建的 docker volume create agent-x-cicd
# 启动所有服务
docker-compose up -d
# 查看运行状态
docker-compose ps
# 查看日志
docker-compose logs -f [服务名]