MonkeAgent

Managed hosting for OpenClaw with MonkeDAO branding. Two halves of the same product: a customer-facing storefront — landing, signup wizard, tenant dashboard — and the multi-tenant infrastructure that quietly runs each customer’s agent on its own isolated instance.

The hosting layer

Every customer gets their own private container on a shared host. Full isolation, no shared state, with disk caps so one noisy tenant can’t crowd out another. The whole thing is operated from a single script: create a customer, rotate a secret, resize a quota, tail logs. Secrets — model API keys, bot tokens — stay locked to each tenant’s environment, never written into config files or baked into images.

The onboarding flow

Customer creation is a single command. It allocates the tenant’s resources, generates their keys, writes their environment, and runs a first-boot routine that wires up the AI provider and the messaging channel without any prompts. Subsequent boots skip onboarding entirely and go straight to running. 40+ AI providers supported (Anthropic, OpenAI, Gemini, Mistral, OpenRouter, DeepSeek, …); each tenant’s bot only accepts messages from people they’ve explicitly allowed.

The storefront

The Next.js side is the front door: hero, features, pricing, a four-step signup wizard (account → bot → AI provider → launch), and a dashboard for status, integrations, and activity. It’s a thin layer over the same operator primitives the script exposes — not a parallel system to keep in sync.