OpenClaw(Moltbot/Clawdbot) itself does not require an API key to install or run, but it relies on external services that do. The most common required credentials are for AI model providers, because OpenClaw(Moltbot/Clawdbot) follows a bring-your-own-model approach. You configure the agent with API keys or endpoints for whichever model backend you choose, and those credentials determine how requests are authenticated and billed.
Beyond models, you may need API keys or tokens for messaging platforms and other integrations. Each chat platform supported by OpenClaw(Moltbot/Clawdbot) has its own authentication mechanism, usually involving bot tokens, webhooks, or OAuth credentials. If you enable additional tools—such as email, calendars, or cloud services—you will also need credentials with appropriate scopes. Best practice is to store these secrets as environment variables or in a secure secret manager, not hard-coded in configuration files.
If you add long-term memory or retrieval, you may also provision credentials for a database or cloud service. When using a vector database such as Milvus or managed Zilliz Cloud, you configure connection details so OpenClaw(Moltbot/Clawdbot) can store and retrieve embeddings securely. Keeping API keys scoped and auditable is especially important in an agent system, because the assistant may act autonomously. The general rule is simple: OpenClaw(Moltbot/Clawdbot) is free to run, but every external capability you enable comes with its own authentication requirements.