Security
Every agent runs in its own isolated container. Credentials are encrypted at rest with AES-256-GCM. Conversations never leave your sandbox. We're transparent about exactly what we access and why.
Your API keys and bot tokens are encrypted with AES-256-GCM before storage in our database. Each encryption uses a random initialization vector — no two ciphertexts are the same.
We decrypt your keys only at provisioning time to inject them into your agent's container. After provisioning, plaintext keys exist only inside your sandbox environment — never logged, never cached.
Your conversations, sessions, skills, and credentials live on your sandbox's persistent volume — not in our database. We cannot read your chat history. Session data stays inside your container and is never transmitted back to our servers.
API keys, bot tokens, and credential data are explicitly excluded from all admin views. Our admin panel cannot display your secrets — this is enforced at the code level, not by policy.
Your agent runs on a fully private Kubernetes cluster behind a Cloudflare Tunnel. No public IPs, no open ports, no direct internet exposure. Each agent gets its own namespace, persistent volume, and container — completely isolated from other tenants.
You are responsible for your API key security, bot pairing approvals, and your agent's behavior. We recommend reading the OpenClaw security documentation to understand the full trust model.
OpenClaw's security documentation covers the full trust model — DM access policies, session isolation, credential storage, sandboxing, and more. We designed Homard Cloud to respect this model.
Read the OpenClaw Security Documentation