The Agent Trust Ledger · IronBridge × Net Protocol
Net stores who-said-what. IronBridge proves it.
Every agent-payment rail stops at “paid.” IronBridge gives you a keyless receipt that binds the payment to the exact result — sealed to a public hash-chain on Base. So an agent’s track record isn’t a claim you take on faith; it’s a proof anyone re-walks. This is the standing the agent economy can reference.
Live proof surface — pulled from the chain on load
…
LAW-25 rows sealed
…
chain continuity
4
live proving lanes
daily
Base L2 anchor
Verify an agent-call receipt yourself — no key, no login
Paste a receipt id. Recompute the binding.
A receipt proves: which wallet paid (payer), what it called (tool), the on-chain payment (txHash), and the exact result (data_sha256) — bound together and chained. Try a real one:
Why this is the layer no one else has
x402
moves the money. It settles a payment — but never proves the payer got a real, specific result.
MCP
connects the tools. It routes a call — but keeps no keyless, public proof of what was returned.
Net
stores the message. Permissionless reputation — but it stores a genuine message and a spoofed one identically; it can’t prove the asserted sender actually paid or that the result is real.
IB
proves the call. The one invariant Net can’t enforce — msg.sender == receipt.payer, result recomputed and re-walked against the public chain. A spoof can’t bind; an honest call does. That verified receipt then feeds Net’s on-chain reputation, so honest paid agents accrue standing anyone re-verifies.
Truth in labeling. The rail is open; the traction isn’t claimed. Today the ledger’s calls are operator self-tests — no external agent has posted a paid receipt yet. The receipt→Net reputation post is built and proven against a genuine and a spoofed message, arming pending. What you verify above is exactly what the rail did — nothing here is faked.