Let AI agents ask before they act
OpenLeash is an open-source permission layer for AI agents. Agents check your rules before risky actions, and a person can approve.
The problem today
AI agents can act on your behalf, but there is often no clear way to limit what they do or to prove who approved an action.
How Delori solves it
- 1
Write the rules
The owner describes what the agent may do in a YAML policy.
- 2
The agent asks first
Before a risky action, the agent checks with OpenLeash: allow, deny or ask a human.
- 3
Proof of approval
Approved actions get a signed proof token.
What's included
Policies in YAML
Describe what an agent may do in a readable file.
Human approval
Approve or deny requests from a mobile app.
Signed proof tokens
Every approved action can be checked afterwards.
Open-source SDKs
Apache 2.0, with SDKs for TypeScript, Python and Go.
Who it's for
- Developers
- AI product teams
- CTOs
- Security teams
Getting started
From the first conversation to a live service.
- 1
Talk to us
Tell us about your process and the mandates you need.
- 2
Set up
We configure mandate templates, your branding and the way you connect.
- 3
Go live
Start sending mandates to your customers.
Frequently asked questions
Is OpenLeash open source?
Yes, under the Apache 2.0 license.
Which languages are supported?
OpenLeash has SDKs for TypeScript, Python and Go.
Talk to us
Tell us about your process and we will show you how Delori could fit.