Frequently asked questions

OpenClaw is an open-source autonomous AI agent that runs as a 24/7 background process and connects through messaging apps like WhatsApp, Telegram, Slack, and Signal. We host it for you on a dedicated VPS so you don't have to set anything up.

No. Runspry is infrastructure only — we provision, run, and back up the VPS, but we don't help with configuring OpenClaw, debugging your agent's prompts, wiring it up to Telegram/Signal/etc., or writing custom plugins. The OpenClaw project documentation and community are the right place for that. If you need a managed service with white-glove setup and human support, you're better off with a different provider.

OpenClaw runs arbitrary shell commands and browser automation. Sharing a Linux kernel across customers is unsafe — a container escape would expose other customers' data. The cost difference between dedicated and shared is small, and isolation is non-negotiable for an agent that has shell access.

Yes. Paste it in the dashboard at /byo_llm_keys; we encrypt it at rest and forward your agent's requests to the upstream provider directly using your key. We don't meter or mark up BYO requests — you pay the upstream provider.

Your agent's LLM calls return HTTP 402 (insufficient credits). The agent itself keeps running on your VPS — only the LLM calls fail until you top up. You can never owe us money.

Files and OpenClaw memory live on your VPS, in /data/agent. We snapshot it daily for backup; snapshots stay in encrypted backup storage with 7-day retention. We never read the contents.

Yes. Cancel from /billing. Your VPS continues running through the end of the current period; we then take a final snapshot, retain it for 30 days, and destroy the server.

Not yet. The cheapest path is to sign up, top up the smallest credit pack ($10), activate Standard for one month, and cancel before renewal. You'll have spent at most $25 to fully evaluate.

Onboarding wizards for each are coming. For now, SSH access via the dashboard plus the OpenClaw docs cover every supported platform.

Yes. Issue a token at /api_tokens and call /api/v1/* — full lifecycle for instances, snapshots, credits, and subscriptions. The API mirrors what the dashboard does, scoped per token.

We're building a partner program (Pressable-style). Contact us if interested.

Per-customer dedicated VPS, hardened Ubuntu 24.04 base image, UFW + fail2ban + automatic security updates, SSH only via Tailscale, audit log on every billing/lifecycle event, encrypted-at-rest API keys, rate limits on every public endpoint, and prepay-only billing so abuse can't drain a payment method.