Skip to main content
OpenFang is configured via a single TOML file located at ~/.openfang/config.toml. This file controls all aspects of the system including models, channels, security, memory, and networking.

​
Quick Start

Create your configuration file: 
cp ~/.openfang/config.toml.example ~/.openfang/config.toml
Or initialize with the setup wizard: 
openfang init

​
Configuration File Location

  • Default path: ~/.openfang/config.toml
  • Custom path: Set via OPENFANG_CONFIG environment variable
  • Example file: Included in installation at ~/.openfang/config.toml.example

​
Core Configuration Sections

Models

Configure LLM providers, model routing, and fallback chains

Providers

Setup API keys and endpoints for 27 LLM providers

Channels

Connect to 40+ messaging platforms and configure behavior

Security

Enable authentication, rate limiting, and security features

​
Minimal Configuration

The absolute minimum configuration requires only a default model: 
[default_model]
provider = "anthropic"
model = "claude-sonnet-4-20250514"
api_key_env = "ANTHROPIC_API_KEY"
Set the API key as an environment variable: 
export ANTHROPIC_API_KEY="sk-ant-..."
openfang start

​
Full Configuration Example

# API server settings
api_key = ""                               # Set to enable Bearer auth (recommended)
api_listen = "127.0.0.1:50051"             # HTTP API bind address (use 0.0.0.0 for public)

[default_model]
provider = "anthropic"                    # "anthropic", "gemini", "openai", "groq", "ollama", etc.
model = "claude-sonnet-4-20250514"        # Model identifier
api_key_env = "ANTHROPIC_API_KEY"         # Environment variable holding API key
# base_url = "https://api.anthropic.com"  # Optional: override API endpoint

[memory]
decay_rate = 0.05                         # Memory confidence decay rate
# sqlite_path = "~/.openfang/data/openfang.db"   # Optional: custom DB path

[network]
listen_addr = "127.0.0.1:4200"           # OFP listen address
# shared_secret = ""                      # Required for P2P authentication

# Session compaction (LLM-based context management)
[compaction]
threshold = 80                          # Compact when messages exceed this count
keep_recent = 20                        # Keep this many recent messages after compaction
max_summary_tokens = 1024               # Max tokens for LLM summary

# Usage tracking display
usage_footer = "Full"                   # "Off", "Tokens", "Cost", or "Full"

# Channel adapters (configure tokens via environment variables)
[telegram]
bot_token_env = "TELEGRAM_BOT_TOKEN"
allowed_users = []                      # Empty = allow all

[discord]
bot_token_env = "DISCORD_BOT_TOKEN"
guild_ids = []                          # Empty = all guilds

[slack]
bot_token_env = "SLACK_BOT_TOKEN"
app_token_env = "SLACK_APP_TOKEN"

# MCP server connections
[[mcp_servers]]
name = "filesystem"
command = "npx"
args = ["-y", "@modelcontextprotocol/server-filesystem", "/tmp"]

​
Environment Variables

OpenFang uses environment variables for sensitive credentials:

​
API Keys

# LLM Providers
export ANTHROPIC_API_KEY="sk-ant-..."
export OPENAI_API_KEY="sk-..."
export GEMINI_API_KEY="..."
export GROQ_API_KEY="gsk_..."

# Channel Adapters
export TELEGRAM_BOT_TOKEN="..."
export DISCORD_BOT_TOKEN="..."
export SLACK_BOT_TOKEN="xoxb-..."
export SLACK_APP_TOKEN="xapp-..."

# Web Services
export BRAVE_API_KEY="..."
export TAVILY_API_KEY="..."
Never hardcode API keys in config.toml. Always use environment variables referenced via api_key_env fields.

​
Hot Reload

OpenFang can automatically reload configuration changes without restarting: 
[reload]
mode = "hybrid"          # "off", "restart", "hot", or "hybrid"
debounce_ms = 500       # Wait time before applying changes

​
Reload Modes

  • off: No automatic reloading (default for production)
  • restart: Full daemon restart on config change
  • hot: Hot-reload safe sections only (channels, skills, heartbeat)
  • hybrid: Hot-reload where possible, flag restart-required otherwise
Some settings like api_listen and security settings require a full restart and cannot be hot-reloaded.

​
Validation

Validate your configuration before starting: 
openfang config validate
View merged configuration (includes defaults): 
openfang config show

​
Configuration Hierarchy

Configuration is loaded with the following precedence (highest to lowest):
  1. Environment variables (for API keys)
  2. Command-line flags (e.g., --api-listen)
  3. config.toml (user configuration)
  4. Built-in defaults (in code)

​
Next Steps

Configure Models

Set up LLM providers and model routing

Setup Providers

Configure all 27 supported LLM providers

Enable Channels

Connect messaging platforms

Secure Your Instance

Enable authentication and security features