cloudflare/inference.js: one Worker (time-inference) at inference.<TIME_VENTURE_DOMAIN> holding VENICE_API_KEY as a secret; venture pages send the same bodies they would send Venice, with an x-app header. Allowlisted routes (chat, music quote/queue/retrieve/complete, image generate), allowlisted models, size caps, and per-person / per-app daily quotas in D1 (the same counter pattern as worker.js). inference.py deploys it; idempotent. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01K6897PWcUEwHiqZMcVnNF6