Safe AI Agent Execution
TOTP-authenticated command gateway that intercepts dangerous shell commands and routes approval to your phone. Prevent LLM agents from executing destructive operations without your consent.
npx -y yespapa initInstall the CLI
YesPaPa works completely offline with any TOTP authenticator. Add the mobile app for push notifications and one-tap approvals.
Run without installing (recommended)
npx -y yespapa initApprove commands from your phone
Agent runs a dangerous command
Commands like rm -rf are intercepted before execution
You get a push notification
See the full command and justification on your phone
One-tap approve or deny
Add a message to give the agent feedback along with your decision

How YesPaPa Protects You
Two-ring architecture: Inner Ring works offline with any TOTP app. Outer Ring adds push notifications and mobile approvals.
Shell Interception
Automatically intercepts dangerous commands like rm -rf before execution
TOTP Authentication
Works with any standard authenticator app — Google Authenticator, Authy, 1Password
Mobile Approvals
One-tap approve or deny from the YesPaPa app with push notifications
Approve with Message
Send feedback to AI agents along with your approval or denial
Tamper Protection
Re-injects shell aliases if an agent removes them. Agents can't remove their own guardrails
Self-Hostable
Run your own remote server with Docker or Supabase for full control
Get Early Access
Join the beta to get the mobile app and help shape the future of safe AI execution.
By signing up, you agree to receive beta testing updates. We won't spam you.