Configuration¶

Vectra QA is configured through environment variables. You can set them in: - .env file (recommended) - Docker Compose environment section - Shell exports

Required Configuration¶

LLM Provider (Choose at least one)¶

Variable	Description	Example
`OPENAI_API_KEY`	OpenAI API key	`sk-...`
`ANTHROPIC_API_KEY`	Anthropic/Claude API key	`sk-ant-...`
`GOOGLE_API_KEY`	Google Gemini API key	`...`
`MINIMAX_API_KEY`	MiniMax API key	`...`
`KIMI_API_KEY`	Kimi/Moonshot API key	`...`

Core Configuration¶

Variable	Default	Description
`MCP_SERVER_URL`	`http://mcp-server:8080`	URL of the MCP tool server
`OBSIDIAN_VAULT_PATH`	`/app/obsidian_vault`	Path to Obsidian vault directory
`COMMAND_CENTER_PORT`	`3000`	Dashboard HTTP port
`MCP_SERVER_PORT`	`8080`	MCP server HTTP port

Chatbot Configuration¶

Variable	Default	Description
`CHATBOT_MODEL`	`anthropic/claude-3-5-sonnet-20241022`	LLM model for chatbot
`CHATBOT_MAX_HISTORY`	`50`	Maximum conversation history messages
`CHATBOT_ENABLE_STREAMING`	`true`	Enable SSE streaming responses

Browser Configuration¶

Variable	Default	Description
`HEADLESS`	`true`	Run Playwright in headless mode
`PLAYWRIGHT_BROWSER`	`chromium`	Browser engine (chromium, firefox, webkit)

Security¶

Variable	Default	Description
`COMMAND_CENTER_SECRET_KEY`	`dev-secret-key`	JWT/secret key (change in production)
`MCP_API_KEY`	(none)	API key for MCP server authentication
`CORS_ORIGINS`	`*`	Allowed CORS origins

Example .env File¶

# LLM Provider (OpenAI example)
OPENAI_API_KEY=sk-your-key-here

# Core Services
MCP_SERVER_URL=http://mcp-server:8080
COMMAND_CENTER_PORT=3000
MCP_SERVER_PORT=8080

# Chatbot
CHATBOT_MODEL=openai/gpt-4o
CHATBOT_MAX_HISTORY=50

# Browser
HEADLESS=true

# Security
COMMAND_CENTER_SECRET_KEY=your-secure-secret-key-change-this

Multiple LLM Providers¶

You can configure multiple providers simultaneously. The system will use them in this priority:

Anthropic (Claude) — Best for complex reasoning
OpenAI (GPT-4o) — Fast and reliable
Google (Gemini) — Good for long context
MiniMax — Cost-effective
Kimi — Chinese-optimized
Local (Ollama) — Privacy-first

Example with multiple providers:

OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
GOOGLE_API_KEY=...
LOCAL_LLM_BASE_URL=http://localhost:11434/v1

Docker Compose Configuration¶

Override defaults in docker-compose.yml:

services:
  command-center:
    environment:
      - CHATBOT_MODEL=openai/gpt-4o
      - CHATBOT_MAX_HISTORY=100
      - HEADLESS=false  # For debugging

Environment-Specific Configurations¶

Development¶

HEADLESS=false  # See browser windows
LOG_LEVEL=DEBUG
CHATBOT_ENABLE_STREAMING=true

Production¶

HEADLESS=true
LOG_LEVEL=INFO
COMMAND_CENTER_SECRET_KEY=your-strong-secret
MCP_API_KEY=your-mcp-api-key
CORS_ORIGINS=https://yourdomain.com

CI/CD¶

HEADLESS=true
LOG_LEVEL=WARNING
CHATBOT_MODEL=openai/gpt-4o-mini  # Faster, cheaper