proofrelay

Interesting. One pattern we’ve run into is that the hardest part of post incident analysis isn’t the action log, it’s reconstructing the state of authority and context at execution time. A defensible execution record usually ends up needing a bundle like: input context, delegated identity/permissions, policy version in force, intended action, actual outcome, and a cryptographic link to the previous step in the workflow. Without sealing that bundle at execution time, you’re left playing mix and match with logs and systems later. This isn’t really practical if you’re trying to produce an audit grade reconstruction of the decision chain.

Specifically the pre-chain window — the inputs the agent saw, the logic it applied, and why it chose to act or hold before any state transition hit the chain.

On-chain proves what happened. I'm trying to understand how people are handling proof of why the decision was made. Curious whether anyone has solved this or is just living with the gap.

AI agents can now call APIs, execute workflows, and move money.

But there’s no standardized way to generate machine-verifiable proof that an action actually occurred.

Today, most systems rely on: - Logs in a database - Screenshots - Internal audit trails

Those don’t travel well between systems.

I built a small API experiment that issues cryptographically sealed execution receipts.

Under the hood: - The payload is canonicalized - Hashed (SHA-256) - Sealed with HMAC - Timestamped - Stored idempotently

Two endpoints:

POST /execute - Accepts structured JSON - Returns a receipt ID + HMAC seal

GET /verify - Recomputes the hash + verifies integrity

This week I recorded a 90-second demo: An agent “hires” a freelancer, verifies the deliverable, and generates a tamper-evident receipt.

https://www.loom.com/share/845adcf05d2e40c6b495e3b9663fcfd0

Curious about feedback from engineers building: - autonomous agents - workflow systems - distributed automation

Does "proof-of-execution" feel like a real primitive to you? Or is this just glorified logging with extra steps?

Would especially appreciate thoughts on the trust model.

Live at proofrelay.app