What Variables Are

Variables are named values that Wolffish stores in config.json under the variables array. They hold API keys, tokens, and secrets that capabilities need at runtime. Configure them in Settings > Variables.

Structure

Each variable has three fields:

Field	Type	Description
`name`	string	Identifier used to reference the variable (e.g., `OPENAI_API_KEY`)
`value`	string	The actual secret or value
`sensitive`	boolean	When true, the value is masked in the UI

{
  "variables": [
    {
      "name": "OPENAI_API_KEY",
      "value": "sk-proj-abc123...",
      "sensitive": true
    },
    {
      "name": "GITHUB_TOKEN",
      "value": "ghp_xxxxxxxxxxxx",
      "sensitive": true
    },
    {
      "name": "BRAVE_API_KEY",
      "value": "BSA-xxxxxxxxxx",
      "sensitive": true
    },
    {
      "name": "NOTION_TOKEN",
      "value": "ntn_xxxxxxxxxx",
      "sensitive": true
    },
    {
      "name": "DEFAULT_BRANCH",
      "value": "main",
      "sensitive": false
    }
  ]
}

Sensitive Flag

When sensitive is true, the variable’s value is displayed as •••••• in the Settings UI. This is a UI-only convenience — it does not mask or exclude the value from the agent.

Mark any credential, token, or API key as sensitive so it’s hidden in the UI. Non-sensitive variables are for non-secret configuration like branch names, project IDs, or environment labels.

How the Agent Uses Variables

Every variable — including ones marked sensitive — appears in plaintext in the system prompt inside a <variables> block on every turn:

- OPENAI_API_KEY = sk-proj-abc123... (sensitive)
- GITHUB_TOKEN = ghp_xxxxxxxxxxxx (sensitive)
- BRAVE_API_KEY = BSA-xxxxxxxxxx (sensitive)
- DEFAULT_BRANCH = main

The (sensitive) label is a hint to the agent not to echo the value back — it does not hide or redact the actual value. This is by design: Variables are the mechanism for giving the LLM credentials it needs to use directly (e.g. passing an API key as a shell command argument). The agent uses them automatically. If a task needs an API key and a matching variable exists, it uses it without asking.

If you store a secret as a Variable, the LLM sees it in plaintext. This is intentional — the agent needs the value to act on it. Use Variables for credentials the agent needs to reference directly. For credentials that plugins handle themselves (Google OAuth, Notion token, etc.), use the integration-specific settings pages instead — those are never sent to the LLM.

Values are stored in plain text in config.json. Wolffish is local-first and your workspace is yours. If you version-control your workspace, add config.json to .gitignore.

Variables vs Environment Variables

Variables are not system environment variables. They differ in important ways:

	Wolffish Variables	System Env Vars
Scope	Wolffish workspace only	Entire OS session
Portable	Yes, travel with the workspace	No, tied to the machine
Managed by	Settings UI or config.json	Shell profile or system settings
Available to	The agent (via system prompt)	All processes

If you need a value available to both Wolffish and other tools on your system, set it as both a Wolffish variable and a system env var. They don’t conflict.

Common Variables

Variable	Used By
`OPENAI_API_KEY`	OpenAI provider fallback
`GITHUB_TOKEN`	GitHub capability (PRs, issues, repos)
`BRAVE_API_KEY`	Web search capability
`NOTION_TOKEN`	Notion integration
`GOOGLE_API_KEY`	Google Workspace integration
`TELEGRAM_BOT_TOKEN`	Telegram channel

Adding a Variable

Via Settings UI
Via config.json

Open Settings > Variables
Click Add Variable
Enter the name, value, and toggle sensitive if needed
Click Save

Edit brain/config.json directly and add an entry to the variables array:

{
  "name": "MY_API_KEY",
  "value": "your-key-here",
  "sensitive": true
}

Wolffish picks up the change on next conversation turn (no restart needed).

Get Started

Core Concepts

Capabilities

Channels

Models

Integrations

Memory & Learning

Configuration

Debugging

Extending Wolffish

Development

Variables

What Variables Are

Structure

Sensitive Flag

How the Agent Uses Variables

Variables vs Environment Variables

Common Variables

Adding a Variable

​What Variables Are

​Structure

​Sensitive Flag

​How the Agent Uses Variables

​Variables vs Environment Variables

​Common Variables

​Adding a Variable

What Variables Are

Structure

Sensitive Flag

How the Agent Uses Variables

Variables vs Environment Variables

Common Variables

Adding a Variable