ShellBot gives each customer a dedicated agent environment, with scoped permissions, encrypted access, and approval before any sensitive action. Use it from WhatsApp, Telegram, or Signal without putting your data in a shared assistant.
Dedicated environment
Each customer gets an isolated agent machine, not a shared assistant pool.
Encrypted access
Keys and credentials stay protected with scoped access and encrypted storage.
Human approval
Sensitive actions require explicit confirmation before the agent proceeds.
Your agent asks before acting, logs what it does, and can be stopped instantly. You stay in charge even when the request starts from a messaging app.
Your agent asks before sending emails, deleting files, or modifying data. Nothing destructive happens without your explicit approval.
Automatic limits prevent runaway costs. If something loops, we catch it before it drains your credits.
Prompt injection defenses detect and block known attack patterns. Multiple layers of protection help keep your agent's safety rules intact.
Your agent only accesses what you allow. Email read doesn't mean email send. Calendar view doesn't mean calendar edit.
Actions your agent takes are logged and reviewable from the dashboard. Track what happened and when.
One click to pause or stop your agent immediately. If anything looks off, you're always one tap away from full control.
Four barriers between the internet and your private data.
Two-factor authentication: password + phone
All traffic encrypted to the same standard as online banking
Your machine is hidden behind a firewall. No public access
A dedicated computer just for you. No one else's data touches it
Every part of ShellBot is designed to protect your data and your privacy.
Two-factor login means nobody gets in without your password and your phone.
The keys that connect to your machine are encrypted and never visible to anyone, not even our team.
Your agent runs behind a firewall on a private network. There's no way to reach it directly from the internet.
Your AI agent runs in an isolated container. It can't affect other systems, and every plugin is scanned for safety.
The AI model itself runs at the model provider — same infrastructure that powers ChatGPT or Claude. Your conversations are processed there. Your files, agent configuration, and operational logs stay on your private server.
When you connect Gmail, Slack, or Calendar, those credentials are encrypted and stored separately from your agent's machine — never on the server that runs your agent.
SOC 2
Controls mapped to Security & Availability criteria. Not yet certified.
GDPR
Data encryption, access controls, minimal collection. EU data centers.
PCI DSS
Stripe handles all card data. ShellBot does not process card numbers.
ISO 27001
Security controls informed by ISO 27001 framework. Not yet certified.
Your ShellBot only runs verified plugins. Every extension is automatically scanned before it reaches your machine.
OpenClaw
VirusTotal Sentinel
OpenClaw partners with VirusTotal (the world's leading threat intelligence platform) to scan every plugin. Each one is analyzed by multiple security engines and AI-powered code analysis before it's available to install.
Read about the OpenClaw & VirusTotal partnership