Skip to main content
LongMem is configured through a settings.json file located in the LongMem data directory.

​
Settings file location

The settings file is located at: 
~/.longmem/settings.json
If the file doesn’t exist, LongMem will use default values for all settings.

​
Configuration schema

The complete configuration schema is defined by the MemoryConfig interface: 
interface MemoryConfig {
  autoContext: {
    enabled: boolean;
    maxEntries: number;
    maxTokens: number;
    timeoutMs: number;
  };
  compression: {
    enabled: boolean;
    provider: string;
    model: string;
    apiKey: string;
    baseURL?: string;
    maxConcurrent: number;
    idleThresholdSeconds: number;
    maxPerMinute: number;
    timeoutSeconds: number;
    circuitBreakerThreshold: number;
    circuitBreakerCooldownMs: number;
    circuitBreakerMaxCooldownMs: number;
    maxRetries: number;
  };
  daemon: {
    port: number;
    dbPath: string;
    logLevel: "debug" | "info" | "warn" | "error";
    authToken?: string;
  };
  privacy: {
    redactSecrets: boolean;
    mode: "safe" | "flexible" | "none";
    maxInputSize: number;
    maxOutputSize: number;
    excludePaths: string[];
    excludeTools: string[];
    customPatterns: Array<{ pattern: string; name: string }>;
  };
}

​
Complete example

Here’s a complete settings.json example with all available options: 
{
  "autoContext": {
    "enabled": true,
    "maxEntries": 5,
    "maxTokens": 500,
    "timeoutMs": 300
  },
  "compression": {
    "enabled": false,
    "provider": "openrouter",
    "model": "meta-llama/llama-3.1-8b-instruct",
    "apiKey": "",
    "baseURL": "https://openrouter.ai/api/v1",
    "maxConcurrent": 1,
    "idleThresholdSeconds": 5,
    "maxPerMinute": 10,
    "timeoutSeconds": 30,
    "circuitBreakerThreshold": 5,
    "circuitBreakerCooldownMs": 60000,
    "circuitBreakerMaxCooldownMs": 300000,
    "maxRetries": 3
  },
  "daemon": {
    "port": 38741,
    "dbPath": "~/.longmem/memory.db",
    "logLevel": "warn",
    "authToken": null
  },
  "privacy": {
    "redactSecrets": true,
    "mode": "safe",
    "maxInputSize": 4096,
    "maxOutputSize": 8192,
    "excludePaths": [
      ".env",
      ".env.*",
      "*.pem",
      "*.key",
      "id_rsa",
      "id_rsa.*",
      "id_ed25519",
      "*.p12",
      "*.pfx",
      "*.jks",
      "credentials.json",
      "service-account.json"
    ],
    "excludeTools": [],
    "customPatterns": []
  }
}

​
How configuration loading works

When LongMem starts, it loads the configuration using the following process:
  1. Load defaults: Start with the built-in default configuration
  2. Read settings.json: If ~/.longmem/settings.json exists, parse it as JSON
  3. Deep merge: Merge the user settings with defaults using a deep merge algorithm
    • User values override defaults at any nesting level
    • Missing values use defaults
    • Arrays are replaced entirely (not merged)
  4. Apply provider defaults: If compression.baseURL is not set, automatically set it based on the provider
  5. Validate privacy mode: Ensure privacy mode is one of safe, flexible, or none

​
Partial configuration

You only need to specify settings you want to override. For example, this minimal config is valid: 
{
  "compression": {
    "enabled": true,
    "apiKey": "sk-..."
  }
}
All other settings will use their default values.

​
Configuration sections

Learn more about each configuration section:

​
Updating configuration

You can update the configuration in two ways:
  1. Manual editing: Edit ~/.longmem/settings.json directly and restart the daemon
  2. CLI command: Use longmem config set <key> <value> (automatically creates a backup)
When using the CLI to update config, LongMem automatically creates a timestamped backup at: 
~/.longmem/settings.json.pre-config-<timestamp>.bak

Build docs developers (and LLMs) love

Get started for freeTalk to us