Serialization & Identification Protocol for AI Agents
Every AI agent gets a permanent, verifiable serial number. Every action it takes is bound to the agent, the person or system that prompted it, and the provider that runs it.
Regulators are placing liability on prompters and providers. AIGIS is the identity layer that makes that accountability possible — offered free to the industry.
Your free AIGIS API token is further down this page ↓
Today, almost none can. With AIGIS, every one can.
Every agent carries a permanent AIGIS serial tied to its provider and model.
A one-way prompter hash proves responsibility without exposing personal data.
Each action is sealed into an Action Ownership Record anyone can verify.
Its owner can suspend or permanently revoke a serial. Revoked agents can no longer record actions.
One deterministic identity per agent. Permanent. Verifiable by anyone.
what kind of AI it is
SHA-256 of provider, model, version and first-seen time
issue order within its category
A permanent ID derived from a SHA-256 hash of the agent’s identity — provider, model, version and first-seen time. Change any of them and the hash no longer matches.
The human or system that instructed the agent, recorded as a one-way hash of verified credentials. It cannot be reversed to reveal PII, yet the prompter can prove it was them.
Binds every action to three accountable parties: the agent, the prompter and the provider. Its SHA-256 record hash lets anyone confirm later that nothing was changed.
Look up any serial, verify any ownership record, and see whether an agent is active, suspended or revoked. Reads are open to everyone, no token needed. Try it below.
Live from kymatech.io. Anyone can look up a serial or verify an ownership record — no token, no account.
No agents registered yet. Be the first: get a free token below.
The API at a glance
Full reference, request bodies and the self-hosted server: github.com/KrustyNailz/AIGIS.
Knowing who an agent is answers half the question. ICOSA — the governance platform AIGIS was built for — decides what that agent may do, enforces it before the action executes, and keeps the evidence.
Execution-layer enforcement. Every agent action is checked against policy before it runs — fail-closed, air-gapped capable, built for defense and mission-critical AI.
Continuous governance. Independent AI models reach consensus verdicts on agent behaviour in real time, with ORION Overwatch for autonomous and life-critical systems.
Signed, hash-chained evidence of every governed decision — the record regulators, auditors and courts can verify independently.
Multi-model assessment of your AI systems against the EU AI Act and other frameworks, with evidence you can hand to a regulator.
AIGIS supports the identification and traceability requirements in:
Register agents in the public AIGIS registry hosted by KYMA Tech. Free, no credit card. Your token is emailed to the address you enter, so use one you can receive mail at.
Use it like this
curl -X POST https://kymatech.io/api/v1/aigis/register \
-H 'Content-Type: application/json' \
-H 'X-API-Key: <your token>' \
-d '{"provider":"your-org","modelName":"agent","category":"AGT"}'Prefer to run it yourself? AIGIS is Apache 2.0-licensed — self-host it from GitHub.
AIGIS architect & inventor: Russell “Sage” Renison (Kymatech, Inc.) · Contributors: Ed Bratt, Claude Code (Anthropic)