Configuration¶

LEGIONHERCULES can be customized through configuration files and environment variables.

Configuration File¶

Location¶

LEGIONHERCULES looks for configuration in:

Current directory: ./.legionhercules/config.yaml
User home: ~/.legionhercules/config.yaml

Creating Configuration¶

legionhercules config --init

This creates a default configuration file.

Viewing Configuration¶

legionhercules config --show

Configuration Options¶

LLM Settings¶

llm:
  provider: ollama
  model: llama3.2
  base_url: http://localhost:11434
  temperature: 0.7
  max_tokens: 4096
  timeout: 120.0

Option	Description	Default
provider	LLM provider (currently only ollama)	ollama
model	Model name	llama3.2
base_url	Ollama server URL	http://localhost:11434
temperature	Creativity (0.0-2.0)	0.7
max_tokens	Maximum tokens per response	4096
timeout	Request timeout in seconds	120.0

Orchestrator Settings¶

orchestrator:
  max_concurrent_tasks: 5
  task_timeout: 300.0
  retry_attempts: 3
  enable_parallel: true

Option	Description	Default
max_concurrent_tasks	Maximum parallel tasks	5
task_timeout	Task timeout in seconds	300.0
retry_attempts	Number of retries on failure	3
enable_parallel	Enable parallel execution	true

Logging Settings¶

logging:
  level: INFO
  file: null
  use_rich: true

Option	Description	Default
level	Log level (DEBUG, INFO, WARNING, ERROR)	INFO
file	Log file path (null for none)	null
use_rich	Use Rich for console output	true

Agent Definitions¶

agents:
  - name: default
    description: Default general-purpose agent
    system_prompt: You are a helpful AI assistant.
    tools:
      - file_read
      - file_write
      - file_edit
      - bash
      - web_search
    max_iterations: 10
    timeout_seconds: 120.0

Environment Variables¶

Override configuration with environment variables:

export LEGIONHERCULES_LLM_MODEL=llama3.1:8b
export LEGIONHERCULES_LLM_TEMPERATURE=0.5
export LEGIONHERCULES_LLM_MAX_TOKENS=2048
export LEGIONHERCULES_ORCHESTRATOR_MAX_CONCURRENT_TASKS=10
export LEGIONHERCULES_LOGGING_LEVEL=DEBUG

Variable naming: LEGIONHERCULES_<section>_<option>

Example Configurations¶

Minimal Config¶

llm:
  model: llama3.2

Development Config¶

llm:
  model: codellama
  temperature: 0.3

logging:
  level: DEBUG
  file: ~/.legionhercules/debug.log

orchestrator:
  max_concurrent_tasks: 10

Production Config¶

llm:
  model: llama3.1:8b
  temperature: 0.5
  max_tokens: 2048

logging:
  level: WARNING

orchestrator:
  max_concurrent_tasks: 3
  retry_attempts: 5

Per-Project Configuration¶

Create a local config in your project:

mkdir -p .legionhercules
cat > .legionhercules/config.yaml << 'EOF'
llm:
  model: codellama

agents:
  - name: project_agent
    description: Specialized for this project
    system_prompt: |
      You are working on a Python web application.
      Follow the project's coding standards.
    tools:
      - file_read
      - file_write
      - bash
EOF

LEGIONHERCULES will use this config when run from the project directory.