ABOUT

KYA-OS is an engineering task force.
We define primitives for AI identity.
Open standards for the agentic era.

Philosophy

AI agents are becoming first-class actors on the internet. They read our emails, manage our calendars, write our code. They act on our behalf.

This requires trust infrastructure that doesn't exist yet. Who is this agent? What is it authorized to do? How do we limit the blast radius when things go wrong?

KYA-OS defines the primitives to answer these questions. Self-sovereign identity. Bounded delegation. Cryptographic proof. Enforcement at every surface.

Primitives

The foundation is protocol-agnostic:

Identity    →  Unforgeable proof of who
Credentials →  Verifiable claims of capabilities
Delegation  →  Human-issued authority without keys
Consent     →  Signed approval to act
Proof       →  Every action leaves a trace
Audit       →  Who did what, and when

Enforcement Surfaces

The primitives get applied to different surfaces:

KYA-OS Primitives Identity · Credentials · Delegation · Consent · Proof · Audit

MCP-I Tools

Compute Sandbox

Spend Budget

Same primitives. Different attack surfaces. Same trust model.

Not a Product

KYA-OS is not a company. Not a product. Not a service.

It's an open specification with reference implementations. Anyone can implement it. Anyone can extend it. The goal is industry-wide adoption of sound identity practices.

We believe AI can only reach its full potential when trust is cryptographically verifiable and blast radius is bounded by design.

Approach

1. Define primitives (not products)
2. Write specifications (not marketing)
3. Build reference implementations
4. Battle-test in production
5. Share findings with the industry
6. Iterate based on real-world feedback

Get Involved

The specifications are open.
The implementations are open source.
Contributions are welcome.

GitHub · MCP-I Spec

KYA-OS is maintained by engineers who care deeply about
building the identity layer for AI the right way.