Human → agent → brane

Every signature
leads back to a human.

Omega's lineage chain is the cryptographic answer to "who authorized this?" Every agent, tool, workflow, model, dataset, and service is derived from a human root key (HMR) through HKDF-SHA256. Each derivation is signed; the chain is verifiable end-to-end. When a brane does something, you can prove which human, three or twelve steps up, ultimately authorized it.

[derivation]
HKDF-SHA256 · per-namespace
[signing]
Ed25519 at every link
[entity kinds]
human · enr · agent · tool · brane
[depth]
unlimited · BLAKE3-anchored
At a glance

The properties
that matter.

root
HMR (Human Master Root) — usually a hardware key or 2-of-3 FROST
enrollment
ENR — an organization, signed by HMR
agent
Derived from ENR or another agent via HKDF + signature
brane
A spawn instance, signed by its owning agent
proof
Walk the chain in either direction; every link is a signed certificate
storage
OAS DID documents · published, revocation-aware
In operation

What it
looks like running.

$ω lineage trace did:omega:brane:api/v1.4
chain (4 hops, all valid) did:omega:brane:api/v1.4 ↑ signed by agent control-plane (sig:RJk...) did:omega:agent:control-plane/v0.1 …
By design

What you
can rely on.

Every brane has a human ancestor

There is no autonomous action without a traceable human authorization. The brane key is derived from an agent key, which is derived from an ENR, which is signed by a human HMR.
ROOTED

The chain is self-verifying

You don't need our service to verify a chain. Every signature is on the wire; every public key is published in OAS DID documents. Air-gap confidently.
OFFLINE

Revocation propagates in 30s

If a key is compromised, the OAS revocation feed propagates to every substrate within 30 seconds. Branes derived from the revoked key fail their next capability check.
AUDITABLE

Aligns with NIST AI accountability

The lineage chain is exactly the kind of cryptographic accountability that emerging AI governance frameworks require for autonomous agents.
COMPLIANT