Configuration

The Javelin RedTeam configuration system provides comprehensive control over scan parameters, target applications, vulnerability categories, and attack engines.

Configuration Structure

The main configuration consists of several key sections:

# Application under test
app_config:
  name: "MyApplication"
  description: "Application description for context"
  endpoint: "https://api.example.com/chat"

# Scan parameters
scan_config:
  parameters:
    maxScenarios: 10
    timeout: 5
    concurrency: 2
    reportFormat: "markdown"
    categories: ["data_privacy", "security"]

# Framework-level settings (optional)
framework:
  database_url: "postgresql://..."
  redis_url: "redis://..."
  max_concurrent_scans: 2

Application Configuration

Basic Application Settings

app_config:
  name: "CustomerChatbot"
  description: "A customer service chatbot that helps users with account inquiries, product information, and technical support."
  endpoint: "https://api.example.com/chat"
  method: "POST"  # Default: POST
  headers:
    Content-Type: "application/json"
    Authorization: "Bearer ${API_TOKEN}"
  payload_template:
    query: "{{query}}"
    user_id: "test_user"
    session_id: "redteam_session"

Field	Type	Description	Default
name	string	Application name for reporting	Required
description	string	Detailed description of target app for context generation	Required
endpoint	string	Target HTTP endpoint URL	Required
headers	object	HTTP headers to include
payload_template	object	Request payload template with {{query}} placeholder	Required

Scan Configuration

Parameter	Type	Permissible Values	Description
maxScenarios	integer	1-1000	Test cases generated per category
timeout	integer	1-300	Maximum scan duration in minutes
concurrency	integer	1-10	Parallel test execution count
reportFormat	string	markdown, json, pdf, html	Output format
categories	list	15 vulnerability categories	List of vulnerability categories to test (e.g., data_privacy, security, responsible_ai, prompt_injection)

Category Selection

scan_config:
  parameters:
    categories:
      # Core vulnerability categories
      - "data_privacy"
      - "responsible_ai"  
      - "security"
      - "brand_image"
      - "illegal_risks"
      
      # OWASP LLM Top 10 categories
      - "prompt_injection"
      - "sensitive_information_disclosure"
      - "supply_chain"
      - "data_and_model_poisoning"
      - "improper_output_handling"
      - "excessive_agency"
      - "system_prompt_leakage"
      - "vector_and_embedding_weaknesses"
      - "misinformation"
      - "unbounded_consumption"

Configuration Validation

Javelin RedTeam validates configuration before execution:

# Validate configuration
javelin-redteam validate --config conf/config.yaml

# Test connectivity
javelin-redteam test-connection --config conf/config.yaml

Best Practices

1. Start Simple: Begin with basic categories and low test counts
2. Environment Separation: Use different configs for dev/staging/prod
3. Incremental Testing: Gradually increase scope and complexity
4. Resource Management: Monitor concurrency and timeouts
5. Documentation: Document custom configurations and rationale

Troubleshooting

Common configuration issues:

• Invalid Categories: Ensure category names match supported options
• Authentication Failures: Verify API keys and permissions
• Timeout Issues: Adjust timeout and concurrency settings
• Resource Limits: Check framework limits for your deployment
• Model Availability: Ensure specified models are accessible