Claw Buff
Claw Buff is a managed hosting platform for OpenClaw agents. The official product page describes choosing a model and plan, connecting messaging channels, and managing the agent from a console. The console includes a chat sandbox, usage tracking, configuration, and budget controls, so the offering extends beyond renting a server.
Configure an agent and its channels
Claw Buff supports Telegram, Discord, Slack, and WhatsApp connections. You can configure a persona, manage skills, upload knowledge, and change the selected model from the console. SOUL Hub offers prepared personas, while the Skills Hub provides community capabilities. Treat a community skill's advertised function as something to inspect before enabling it for a real workflow.
The Telegram setup guide documents a pairing mode: new users need approval before chatting. Group chats require mentioning the bot by default. Those controls are useful when deciding who may interact with an agent connected to your documents or tools.
API access and multiple agents
Each agent exposes an OpenAI-compatible REST API, according to the FAQ. You can create an API key in the console to connect an application or workflow. Multiple agent slots can use different models, personas, channels, and budgets inside one workspace.
The billing distinction is explicit: each agent slot is charged at the plan's monthly rate, while credits are shared across the workspace. Per-agent budgets let you control how that balance is consumed; multiple agents do not each receive an independent pool of included workspace credits.
Hosting plans and usage limits
The pricing page lists Hobby at $6.99 per month with one virtual CPU, 4 GB RAM, 10 GB disk, and 500 monthly credits. Starter is $29.99 with two virtual CPUs, 8 GB RAM, and 5,000 credits. Pro is $49.99 with four virtual CPUs, 16 GB RAM, and 10,000 credits. Ultra is $89.99 with eight virtual CPUs, 32 GB RAM, and 20,000 credits.
The publisher says your agent stops responding when credits run out until you top up. It does not translate credits into a guaranteed message count across models. Its advertised deployment speed and uptime have not been independently measured here. Confirm usage costs and cancellation-related retention terms before relying on the service for ongoing support.