Why Monitor Billing?
SanctumOS uses Venice AI as a primary inference provider. Venice differentiates itself from OpenAI and Anthropic with a privacy-first architecture: minimal retention of prompts and responses, support for open-weight models, and tiered privacy modes. It's a natural fit for a self-hosted agent stack that treats user data as sovereign.
But privacy-first doesn't mean free. If your agent is burning Venice API credits, it should be able to check its own balance — and you should get an alert before it runs dry.
What It Does
sanctumos/sanctum-venice-billing-monitor ships as both a standalone cron script and an SMCP plugin:
- As a cron job: Polls the Venice
/billing/balanceendpoint on a schedule and logs the result. - As an SMCP tool: Your agent can call
billing_checkto see its current balance mid-conversation.
The output includes a formatted billing block, optional error fields, and local venv documentation for isolated installation.
It's a small plugin — maybe 200 lines of Python — but it illustrates a pattern that every SanctumOS deployment will need: agents that can monitor their own resource consumption.