Klent – one click between your AI agent and a prod blast radius
Human-in-the-loop approval for destructive agent actions before they hit production.

One-click approval without dashboards or auth—genuinely fills the human-in-loop gap.
DevOps engineers, AI agent builders, automation developers needing human gates
I built Ottr to make this a primitive.
Ottr gives you a signed, expiring approval link you can generate from any script. When the link is clicked, the process continues. No accounts. No UI.
Flow:
• Script or agent requests approval → gets a one-time URL
• URL is sent to a human (email, Slack, SMS, etc.)
• Human clicks → script resumes (via polling)
This is early and deliberately minimal. I’m trying to validate whether “approval links” are a useful building block, or if people already solve this in a better way.
Human-in-the-loop approval for destructive agent actions before they hit production.
Ed25519-signed audit trail for AI actions when agents move money.
Approval gates bind to server functions the model can't bypass.
No auto-approval setting exists at all — hard-coded safety constraint when Cursor and Claude can delete prod.
Cryptographic proof of human approval for agent actions—solves a real gap in agent safety architecture.
Human-in-the-loop approval for AI agents via Telegram before risky ops.