This guide will help you start all components of the AstronAgent project in the correct order, including authentication, knowledge base, and core services.
The AstronAgent project consists of the following three main components:
- Casdoor - Identity authentication and single sign-on service (required component, provides SSO functionality)
- RagFlow - Knowledge base and document retrieval service (optional component, deploy as needed)
- AstronAgent - Core business service cluster (required component)
Agent System Requirements
- CPU >= 2 Core
- RAM >= 4 GiB
- Disk >= 50 GB
RAGFlow Requirements
- CPU >= 4 Core
- RAM >= 16 GB
- Disk >= 50 GB
RagFlow is an open-source RAG (Retrieval-Augmented Generation) engine that provides accurate question-answering services using deep document understanding technology.
To start the RagFlow service, run our docker-compose.yml file or docker-compose-macos.yml. Before running the installation commands, please ensure that Docker and Docker Compose are installed on your machine.
# Navigate to RagFlow directory
cd docker/ragflow
# Add executable permissions to all sh files
chmod +x *.sh
# Start RagFlow service (including all dependencies)
docker compose up -d
# Check service status
docker compose ps
# View service logs
docker compose logs -f ragflowAccess Address:
- RagFlow Web Interface: http://localhost:18080
Model Configuration Steps:
- Click on your avatar to enter the Model Providers page, select Add Model, fill in the corresponding API address and API Key, and add both Chat model and Embedding model respectively.
- In the upper right corner of the same page, click Set Default Models to set the Chat model and Embedding model added in the first step as defaults.
Important Configuration Notes:
- Elasticsearch is used by default. To use opensearch or infinity, modify the DOC_ENGINE configuration in .env
- GPU acceleration is supported, start with
docker-compose-gpu.yml
Before starting AstronAgent services, you need to configure the relevant connection information.
# Navigate to astronAgent directory
cd docker/astronAgent
# Copy environment variable configuration
cp .env.example .envEdit the docker/astronAgent/.env file to configure RagFlow connection information:
# Navigate to astronAgent directory
cd docker/astronAgent
# Edit environment variable configuration
vim .envKey Configuration Items:
# RAGFlow Configuration
RAGFLOW_BASE_URL=http://localhost:18080
RAGFLOW_API_TOKEN=ragflow-your-api-token-here
RAGFLOW_TIMEOUT=60
RAGFLOW_DEFAULT_GROUP=星辰知识库Obtaining RagFlow API Token:
- Visit RagFlow Web Interface: http://localhost:18080
- Log in and click on your avatar to enter user settings
- Click API to generate an API KEY
- Update the generated API KEY to RAGFLOW_API_TOKEN in the .env file
2.2 Configure iFLYTEK Open Platform APP_ID, API_KEY, and Related Information (Optional, some built-in features require the use of capabilities from the Open Platform)
For documentation, see: https://www.xfyun.cn/doc/platform/quickguide.html
After creating your application, you may need to purchase or claim API authorization service quotas for the corresponding capabilities:
- Spark LLM API: https://xinghuo.xfyun.cn/sparkapi (For the LLM API, you'll need an additional SPARK_API_PASSWORD available on the page : https://console.xfyun.cn/services/bm4)
- Real-time Speech Recognition API: https://console.xfyun.cn/services/rta
- Image Generation API: https://www.xfyun.cn/services/wtop
Edit the docker/astronAgent/.env file and update the relevant environment variables:
PLATFORM_APP_ID=your-app-id
PLATFORM_API_KEY=your-api-key
PLATFORM_API_SECRET=your-api-secret
SPARK_API_PASSWORD=your-api-password
SPARK_RTASR_API_KEY=your-rtasr-api-keyEdit the docker/astronAgent/.env file and update the relevant environment variables:
AI_ABILITY_CHAT_BASE_URL=https://spark-api-open.xf-yun.com/v1
AI_ABILITY_CHAT_MODEL=your-model-id
AI_ABILITY_CHAT_API_KEY=your-api-keySpark RAG cloud service provides two usage methods:
- Use the APP_ID and API_SECRET created on the iFLYTEK Open Platform
- Directly obtain the Spark dataset ID from the web interface, see: xinghuo_rag_tool.html
If you prefer using command-line tools, you can create a dataset with the following cURL command:
# Create Spark RAG dataset
curl -X PUT 'https://chatdoc.xfyun.cn/openapi/v1/dataset/create' \
-H "Accept: application/json" \
-H "appId: your_app_id" \
-H "timestamp: $(date +%s)" \
-H "signature: $(echo -n "$(echo -n "your_app_id$(date +%s)" | md5sum | awk '{print $1}')" | openssl dgst -sha1 -hmac 'your_api_secret' -binary | base64)" \
-F "name=我的数据集"Notes:
- Please replace
your_app_idwith your actual APP ID - Please replace
your_api_secretwith your actual API Secret
After obtaining the dataset ID, please update it in the docker/astronAgent/.env file:
XINGHUO_DATASET_ID=Edit the docker/astronAgent/.env file to configure the AstronAgent service host address:
HOST_BASE_ADDRESS=http://localhostNote:
- If you're using a domain name for access, replace
localhostwith your domain name - Ensure nginx and minio ports are properly exposed
To start AstronAgent services, run our docker-compose-with-auth.yaml file. This file has integrated the Casdoor authentication service through the include mechanism and will automatically start Casdoor.
# Navigate to astronAgent directory
cd docker/astronAgent
# Start all services (including Casdoor)
docker compose -f docker-compose-with-auth.yaml up -dNote:
- Default Casdoor login credentials: username:
admin, password:123
You can create new applications and organizations in Casdoor as needed, and update the configuration information in the .env file (default organization and application already exist).
Obtaining Casdoor Configuration Information:
-
Visit the Casdoor management console: http://localhost:8000
-
Log in with the default administrator account:
admin / 123 -
Create Organization Go to the http://localhost:8000/organizations page, click "Add", fill in the organization name, save and exit.
-
Create Application and Bind Organization Go to the http://localhost:8000/applications page, click "Add".
When creating the application, fill in the following information:
- Name: Custom application name, e.g.,
agent - Redirect URL: Set to the project's callback address. If the Nginx exposed port is
80, usehttp://your-local-ip/callback; if it's another port (e.g.,888), usehttp://your-local-ip:888/callback - Organization: Select the organization name you just created
- Name: Custom application name, e.g.,
-
After saving the application, record the following information and map it to the project configuration items:
| Casdoor Information Item | Example Value | Corresponding .env Configuration Item |
|---|---|---|
| Casdoor Service Address (URL) | http://localhost:8000 |
CONSOLE_CASDOOR_URL=http://localhost:8000 |
| Client ID | your-casdoor-client-id |
CONSOLE_CASDOOR_ID=your-casdoor-client-id |
| Application Name (Name) | your-casdoor-app-name |
CONSOLE_CASDOOR_APP=your-casdoor-app-name |
| Organization Name (Organization) | your-casdoor-org-name |
CONSOLE_CASDOOR_ORG=your-casdoor-org-name |
- Fill in the above configuration information into the project's environment variable file:
# Navigate to astronAgent directory
cd docker/astronAgent
# Edit environment variable configuration
vim .envAdd or update the following configuration items in the .env file:
# Casdoor Configuration
CONSOLE_CASDOOR_URL=http://localhost:8000
CONSOLE_CASDOOR_ID=your-casdoor-client-id
CONSOLE_CASDOOR_APP=your-casdoor-app-name
CONSOLE_CASDOOR_ORG=your-casdoor-org-name- Restart AstronAgent services to apply the new configuration:
docker compose restart console-frontend console-hubAfter startup, you can access the services at the following addresses:
- Casdoor Admin Interface: http://localhost:8000
- RagFlow Web Interface: http://localhost:18080
- Console Frontend (nginx proxy): http://localhost/
- AstronAgent Official Documentation
- Casdoor Official Documentation
- RagFlow Official Documentation
- Docker Compose Official Documentation
If you encounter issues, please:
- Check the log files of related services
- Review the official documentation and troubleshooting guides
- Submit an Issue on the project's GitHub repository
- Contact the technical support team
Note: For first-time deployment, it is recommended to verify all functionalities in a test environment before deploying to production.