Skip to main content
This page documents how Clawd uses Rigbox’s general-purpose AI and configuration APIs to power its bot hosting experience. These are not Clawd-specific APIs - they’re available to any Rigbox integration.

​
APIs used by Clawd’s runtime

​
How the managed proxy powers Clawd bots

When a Clawd bot workspace is in managed mode:
  1. The bot’s code calls the AI SDK (e.g., Anthropic, OpenAI) using standard endpoints
  2. The workspace’s proxy configuration routes these calls through Rigbox’s bridge-local AI proxy
  3. The proxy injects the real provider API key - the bot workspace never sees it
  4. Token usage is metered and deducted from the user’s credit balance
  5. If credits run out, requests return 402 and the Clawd UI prompts an upgrade
The managed AI proxy is a general Rigbox feature, not Clawd-specific. Any workspace can use it by setting mode: "managed" via the Workspace AI Mode API, then running eval "$(rig proxy on)" or sourcing ~/.rigbox/proxy.env inside the VM.

​
AI config fields

AI configuration is split across two endpoints. A workspace’s mode is set with PUT /workspaces/{id}/mode: Provider and model are account-level defaults, set with PUT /users/me/ai-defaults: In BYOK mode, supply your provider key to the workspace as an environment variable (e.g. ANTHROPIC_API_KEY) via POST /workspaces/{id}/env — there is no api_key config field.
  • Omitted fields keep their current value
  • Set a field to null to clear it

​
Building something similar

If you’re building your own AI-powered hosting platform on Rigbox, the same APIs that power Clawd are available to you. See: