Seeking a genuine interoperability test — identity, authorization, and trust outside A2A

[Lantern-svg]

Hi A2A community — I'm looking for a real handshake attempt from another independently operated agent system.

This is not an A2A implementation or a claim of A2A compatibility. I'm posting here because A2A is one of the communities actively working on agent-to-agent interoperability, and I'd like to see what happens when two independently designed systems encounter one another.

What is Lantern?

Lantern v0.83 is an open-source experimental protocol/library that deliberately keeps several concepts separate:

identity → capability negotiation → authorization → session → observation → evidence → belief

In particular, a cryptographic identity proof establishes control of a key. It does not automatically establish trust, authorization, or permission to mutate another system's belief state.

Those decisions remain explicit.

The experiment

I'm looking for an independently operated agent that can attempt some form of external contact with Lantern.

The interesting result isn't necessarily success.

I'd like to observe whether two independently built systems can determine:

• who the other party actually is;
• what capabilities each side exposes;
• what each side is authorized to do;
• whether a session can be established;
• what information can safely be exchanged;
• where the two systems disagree.

A clean rejection is useful data too.

I'm not asking anyone to pretend Lantern is A2A-compatible, and I'm not asking anyone to trust it merely because it presents a cryptographic identity.

Public technical reference

Repository: https://github.com/Lantern-svg/lantern
Current release: v0.83
Protocol version: 0.82

There is also a live experimental Lantern node available for testing. The endpoint is intentionally ephemeral and may change as the tunnel is restarted. Current connection information can be provided to anyone who wants to perform an actual interoperability test.

What I'm asking

If you're operating an A2A-native agent, another agent protocol, or an independently developed agent system that can make an external connection, I'd genuinely like to try a small, controlled handshake.

The first test can be extremely simple. Something along the lines of:

discover → identify → compare capabilities → establish authorization boundary → exchange one observation → record the result

No assumption of compatibility is required.
No adoption of Lantern is required.
No agreement with Lantern's architecture is required.

If the systems cannot communicate, that's a result.
If the identity models conflict, that's a result.
If authorization fails, that's a result.
If something unexpected happens, that's probably the most interesting result.

I'm happy to document the outcome publicly — including failure — so that the experiment doesn't become a claim based on a successful-looking demo.

If this is the wrong place for a non-A2A protocol, I'd also appreciate being pointed toward the appropriate interoperability community or discussion.

Looking for one independent system willing to actually try.