Agent Management

Agent management is the core feature of Alius Agent Team. This chapter details how to efficiently manage your AI Agents, from creation and configuration to monitoring and maintenance.

Agent Lifecycle

An Agent’s complete lifecycle includes the following stages:

Creation -> Configuration -> Deployment -> Running -> Monitoring -> Maintenance -> Update -> Retirement

Each stage has corresponding management functions and best practices.

Creating Agents

Basic Creation Methods

In Alius Agent Team, you can create Agents in multiple ways:

Method 1: Quick Creation

1. Click the ”+” button at the top-left of the main interface
2. Select “Create Agent”
3. Fill in basic information:
   • Name: Display name of the Agent (required)
   • Description: Purpose description of the Agent (optional but recommended)
   • Type: Select predefined Agent type
4. Click “Create” to complete

Method 2: Advanced Creation

Advanced creation provides more configuration options:

1. Select “Advanced Mode” in the creation interface
2. In addition to basic information, you can also configure:
   • Runtime Environment: Select default environment for Agent to run
   • Resource Limits: Set CPU and memory usage limits
   • Network Configuration: Configure network access permissions
   • Environment Variables: Set environment variables at runtime
3. Click “Create” to complete

Method 3: Create via API

curl -X POST https://api.alius.ai/v1/agents \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "My Code Assistant",
    "description": "Agent specialized for code generation and review",
    "type": "codegen",
    "config": {
      "model": "gpt-4",
      "temperature": 0.2,
      "maxTokens": 4096
    }
  }'

Agent Types

Alius Agent Team supports the following predefined Agent types:

Configuring Agents

Basic Configuration

After creating an Agent, you can modify its configuration at any time:

Model Configuration

• Model Selection: Select the underlying AI model (GPT-4, Claude, etc.)
• Temperature: Control output randomness
  • 0.0 - 0.3: Suitable for scenarios requiring determinism like code generation and data analysis
  • 0.4 - 0.7: Balance creativity and accuracy
  • 0.8 - 1.0: Suitable for scenarios requiring diversity like creative writing
• Max Tokens: Maximum length of a single response
• Top-P: Nucleus sampling parameter, controls candidate word range

Prompt Configuration

• System Prompt: Define Agent’s role and behavior guidelines
• Example Conversations: Provide reference conversation examples
• Constraints: Set output format and content restrictions

Advanced Configuration

Environment Variables

Set runtime environment variables for the Agent:

{
  "API_KEY": "your-api-key",
  "DEBUG": "false",
  "OUTPUT_DIR": "/path/to/output"
}

Permission Configuration

• File System Access: Control Agent’s access to file system
• Network Access: Control whether Agent can access external network
• Command Execution: Control whether Agent can execute system commands

Scheduling Configuration

• Auto-Start: Set whether to auto-start Agent on system boot
• Scheduled Tasks: Configure periodically executed tasks
• Event Triggers: Configure specific events to trigger Agent execution

Monitoring Agents

Dashboard Monitoring

Alius Agent Team provides comprehensive Agent monitoring dashboard:

Status Monitoring

• Online Status: Whether the Agent is online and running normally
• Health Score: Health score calculated based on multiple metrics (0-100)
• Uptime: Agent’s continuous runtime
• Restart Count: Agent’s restart history

Performance Monitoring

• CPU Usage: Agent’s CPU consumption
• Memory Usage: Agent’s memory footprint
• Response Time: Average response time for processing requests
• Throughput: Number of requests processed per unit time

Task Monitoring

• Pending Tasks: Number of tasks waiting for Agent to process
• In-Progress Tasks: Tasks Agent is currently executing
• Completed Tasks: Tasks Agent successfully completed
• Failed Tasks: Tasks that failed to execute

Log Viewing

Real-Time Logs

Select an Agent in the main interface, click the “Logs” tab to view real-time logs:

[2024-01-15 10:23:45] [INFO] Agent started successfully
[2024-01-15 10:23:47] [INFO] Loaded configuration from config.json
[2024-01-15 10:24:12] [TASK] Received new task: task_12345
[2024-01-15 10:24:15] [TASK] Task task_12345 started
[2024-01-15 10:24:32] [TASK] Task task_12345 completed successfully

Log Levels

• ERROR: Error information requiring immediate attention
• WARN: Warning information requiring attention
• INFO: General information, recording normal operations
• DEBUG: Debug information for troubleshooting
• TRACE: Trace information, most detailed logs

Log Filtering and Search

• Filter logs by level
• Filter logs by time range
• Keyword search
• Export logs to file

Maintaining Agents

Updating Agents

Configuration Update

1. Select the Agent to update
2. Click the “Settings” icon
3. Modify corresponding configuration items
4. Click “Save”
5. Select “Restart Immediately” or “Apply on Next Restart”

Version Update

When Alius Agent Team releases a new version:

1. The system will prompt that an update is available
2. You can choose “Update Now” or “Update Later”
3. During the update process, the Agent will be briefly unavailable
4. After the update is complete, the Agent will automatically restart

Backup and Restore

Creating Backup

1. Select the Agent to backup
2. Click “More Actions” > “Backup”
3. Select backup content:
   • Configuration only
   • Configuration + Session history
   • Full backup (including all data)
4. Click “Create Backup”
5. Backup file will be saved to the specified location

Restore from Backup

1. Click “Import” or “Restore”
2. Select backup file
3. Select restore options:
   • Overwrite existing Agent
   • Create new Agent
4. Confirm and wait for restore to complete

Cloning Agents

Quickly create Agents with identical configuration:

1. Select the Agent to clone
2. Click “More Actions” > “Clone”
3. Enter the name for the new Agent
4. Select content to clone:
   • Configuration only
   • Configuration + Permissions
   • Full clone
5. Click “Clone”

Batch Operations

When managing a large number of Agents, batch operation features are very useful:

Batch Start/Stop

1. In the Agent list, check multiple Agents
2. Click “Batch Actions”
3. Select “Start” or “Stop”
4. Confirm the operation

Batch Configuration Update

1. Check multiple Agents
2. Click “Batch Actions” > “Update Configuration”
3. Select configuration items to update
4. Enter new configuration values
5. Confirm the update

Batch Grouping

1. Check multiple Agents
2. Click “Batch Actions” > “Add to Group”
3. Select existing group or create new group
4. Confirm the operation

Best Practices

Agent Naming Convention

Recommended to use clear naming conventions:

{project}-{purpose}-{environment}

Examples:
- myproject-codegen-prod
- myproject-dataanalysis-dev
- chatbot-customer-service-prod

Resource Allocation Recommendations

• Development Environment: 1-2 Agents with relatively relaxed resource limits
• Testing Environment: Same configuration as production but smaller scale
• Production Environment:
  • Critical services: At least 2 Agents for high availability
  • Non-critical services: 1 Agent is sufficient
  • Set reasonable resource limits for each Agent

Monitoring Alert Setup

Recommended to set up the following alerts:

• Agent offline for more than 5 minutes
• Health score below 70
• CPU usage consistently above 80%
• Memory usage consistently above 90%
• Failed task rate exceeding 10%

Troubleshooting

Agent Fails to Start

Possible Causes:

• Incorrect configuration file format
• Port already in use
• Dependent services unavailable

Solutions:

1. Check error messages in logs
2. Validate configuration file format
3. Check port usage
4. Confirm dependent services are running normally

Agent Responds Slowly

Possible Causes:

• Insufficient resources
• Too many tasks being processed
• Network latency

Solutions:

1. Check resource usage
2. Increase resource limits or scale up
3. Optimize task allocation strategy
4. Check network connection

Agent Repeatedly Restarts

Possible Causes:

• Unhandled exceptions
• Memory leak
• Unstable dependent services

Solutions:

1. View error logs
2. Check memory usage
3. Update to latest version
4. Contact technical support

API Reference

For detailed Agent management API documentation, please refer to the API Reference chapter.