I built a PoC to bridge MCP and A2A. Here’s what I learned about “interoperability” (and why it’s harder than it looks)

I've been building multi-agent systems for a while, and the biggest headache is always the same: MCP agents can't talk to A2A agents. It's like having two teams in the same company speaking different languages.

So I built a small PoC to explore this. It's a neutral bridge with 3 layers: Discovery, Verification (SHA-256 receipts), and Arbitration. It runs in 60 seconds with zero dependencies.

But here's the honest part: This is not a production-ready solution. It's a starting point. I'm a non-programmer who built this with AI assistance, and I know it has gaps (no Ed25519 signatures yet, no transport layer).

What I learned:

1. Discovery is the hardest part. How do agents find each other across ecosystems? DNS was the answer for the internet. What's the answer for agents?
2. Trust needs a neutral layer. SHA-256 proves integrity, but not identity. You need something more.
3. The "arbitration" gap is real. When AI agents transact, who resolves disputes? This is a legal-tech gap no one has solved.

I'm sharing this because I want to learn from this community. If you've built multi-agent systems, I'd love to hear your thoughts. The code is open source (Apache 2.0), and I've opened 3 issues for the parts I know are weak.

Links in the comments. 👇

Disclaimer: I'm the author of this PoC. Built with AI assistance. Not a production tool.