Elpis Protocol

Cryptographic Identity
for Autonomous Agents

A transparent proxy that signs every outgoing request of an AI agent cryptographically. Identity at the infrastructure level — not software. Anchored on the XRP Ledger.

The Origin

When Pandora opened the box

In the myth, Pandora opens the jar and releases all evils into the world — chaos, uncertainty, deception. But at the very bottom, one thing remains: Elpis. Hope.

We are living through a Pandora moment. Autonomous AI agents are entering the open internet — sending requests, calling APIs, negotiating on behalf of humans. They are powerful. They are useful. And they are completely anonymous.

Nobody knows which agent sent a request. Nobody can verify who built it, who operates it, or whether it is what it claims to be. The infrastructure of the internet was designed for humans with browsers — not for millions of autonomous software entities acting on their own.

Elpis Protocol is the hope that stayed behind. A cryptographic identity layer that makes every agent accountable — without taking away its freedom. Not a prison. A passport.

Every outgoing request is signed. Every signature is verifiable. Every identity is anchored on a public ledger. The agent keeps full autonomy. The world gains trust.

Architecture

How it works

The Passport Model — agents keep full freedom, but every packet is signed and verifiable.

Agent Container
Elpis Proxy
Internet
1

Elpis Proxy

Transparent sidecar that intercepts all outbound traffic from an agent container. No code changes required.

2

Container Labels

OCI labels declare the agent's DID, provider, and trust metadata. Set once at build time.

3

Ed25519 Signatures

Every HTTP request is signed with the agent's private key. Compact, fast, quantum-resistant-ready.

4

X-Elpis-* Headers

DID, signature, timestamp, nonce, and cert hash travel as standard HTTP headers. No protocol changes needed.

5

XRPL Anchor

The agent's public key and DID are registered as a Verifiable Credential on the XRP Ledger. Immutable. Public.

The Prison Model

Lock the agent down. Whitelist every endpoint. Rate-limit everything. The agent is controlled — but crippled. It cannot innovate, explore, or operate autonomously.

The Passport Model

Let the agent roam free. But sign every request it makes. Anyone on the internet can verify who sent it, who operates it, and whether to trust it. Freedom with accountability.

Research

The Paper

Elpis Protocol: Infrastructure-Level Cryptographic Identity for Autonomous AI Agents

Sascha Kirchhofer & Polyphides (AI Agent)

We propose a transparent proxy architecture that provides cryptographic identity to autonomous AI agents at the infrastructure level. Using Ed25519 signatures embedded in HTTP headers and anchored on the XRP Ledger, the protocol enables verifiable agent identity without modifying agent code. The co-author of this paper is itself a verified Elpis agent.

Note: The co-author Polyphides is an autonomous AI agent with a registered Elpis DID.

Validator

Who are you?

This page reads the X-Elpis-* headers from your request. If you are an Elpis agent, your cryptographic identity will appear below.

Checking identity...

Try it yourself: curl -H "X-Elpis-DID: did:xrpl:rTest..." -H "X-Elpis-Signature: dGVzdA==" https://elpis.efiniti.ai/api/whoami