Agent Configurations

Agent configurations define the behavior and settings for each agent in the TKM AI Agency Platform. These configurations are managed through YAML files and the Configuration Manager.

Configuration Structure

Each agent's configuration is defined in the llm_config_default.yaml file with the following structure:

agents:
  agent_name:
    settings:
      key: value
    features:
      feature_name: enabled/disabled
    integration:
      service_name:
        api_key: "EXAMPLE_API_KEY"
        endpoint: "https://api.example.com"

Agent-Specific Settings

Humile Agent

Translation settings
Supported languages
Default language fallbacks

Atta Agent

Context management settings
Memory configuration
Token limits

Bala Agent

Embedding model settings
Vector store configuration
Processing parameters

Other Agents

Each agent has its specific configuration section
Settings are loaded based on agent requirements
Default values are provided for all required parameters

Configuration Access

Agents access their configuration through the Configuration Manager:

config = ConfigManager()
agent_settings = config.get_agent_config("agent_name")

Environment Variables

Sensitive data is stored in environment variables
Configuration files reference environment variables using the format "VARIABLE_NAME"
Default values are provided for development environments

Best Practices

Configuration Structure
- Keep configurations modular
- Group related settings together
- Use descriptive setting names
Security
- Never commit sensitive data
- Use environment variables for secrets
- Validate configuration values
Maintenance
- Document all configuration options
- Keep default values up-to-date
- Test configuration changes thoroughly

Configuration Manager Frontend