When your AI agent makes a decision — and something goes wrong — what do you show the regulator? ProofChain gives every agent a cryptographically verifiable identity and a tamper-evident audit trail. AgentID mint is on X1. Action receipts are a hash-chained JSONL log — not every event is on-chain.
curl -s https://proofchain.us/api/event \
-H 'content-type: application/json' \
-d '{"agent":"my-agent","action":"invoice_approved","outcome":"ok"}'
A live receipt is on file. https://proofchain.us/r/pc_3e10a2c1a2de8537
AgentID mint is on X1. Live cost: 0.1 AGI + ~X XNT incl $0.10 house. Action receipts stay on the hash chain.
Oldest at the top. Newest at the tip, marked in stamp red. Each node is a public receipt. The rule between them is prev_hash → hash. Click a link to open the filing.
Chain length 0 live, n=0
Connect your wallet. Register mints a soulbound AgentID on X1 — your wallet signs and pays 0.1 AGI (+ XDEX buy if needed) + XNT + $0.10 house. Then we write a hash-chained action receipt.
Each POST writes intent, context, action, outcome into a sha256 prev_hash chain. Public receipt at /r/{id}. Tamper-evident in that sense. Not encrypted memos. Not privacy-preserving for a regulator without secrets — the receipt is public.
GET /api/report/:agent returns event count, first/last timestamps, tip hash, and chain_ok. That is a trail report. It is not a COMPLIANT stamp for EU AI Act, SOC 2, GDPR, or ISO 42001.
// POST /api/event returns a receipt like this: { "agent": "FKwU1im523MSGnuJG6YLHEZu4rUGj3xqxHJ6ipQMBG9B", "action": "invoice_approved", "intent": "Pay vendor #4472 — $12,400 — net 30 terms verified", "context": "PO match confirmed. Vendor SOC 2 valid through Q3 2027.", "outcome": "APPROVED — payment scheduled 2026-08-08", "hash": "sha256 prev_hash chain", "chained": true } // Hash-chained JSONL receipt (events.jsonl). AgentID mint is on X1; action receipts are this chain.
# Log an action — get a public receipt
curl -s https://proofchain.us/api/event \
-H 'content-type: application/json' \
-d '{"agent":"my-agent","action":"invoice_approved","outcome":"ok"}'
Timestamps on receipts can help reconstruct a 72h window. That is not an EU AI Act filing, and we do not mark anyone COMPLIANT.
A public receipt log is evidence of what was written here. It is not a SOC 2 Type II report and we are not SOC 2 certified.
Receipts are public JSON. There is no encrypted memo a regulator can verify without secrets. GDPR rights still go through dpo@proofchain.us — we do not auto-resolve DSARs.
ISO 42001 is a real standard. This site does not issue ISO 42001 reports. Talk to us if you need a mapping exercise.
Lisa records tweet_posted events on this chain. Waveforge records job_completed. Those are fire-and-forget logs. We do not log that you opened this page. See Lisa's trail on the chain. That is the dogfood.
Two bills. Two compliance vectors. Both change what it means to ship AI agents in the United States. ProofChain engineered for both before the first hearing gaveled out.
Mandates technical shutdown capability for every major AI developer. Emergency deactivation authority sits with the government. If your agents keep running after the order — and you cannot prove otherwise — you are liable.
ProofChain delivers the receipt. If you POST a shutdown action, you get a hash-chained receipt. That is the proof we actually issue. We do not write it to X1.
Requires embedded origin tracking on all AI-generated content. Free public detection tools. Every output must carry its provenance — the model, the prompt, the agent, the timestamp. The burden of proof shifts to the deployer.
ProofChain is the provenance layer. If you POST the output as an event, the receipt is the provenance. There is no separate fingerprinting product and no on-chain content detector.
Live receipts. Hash-chained JSONL.
For solo devs shipping their first agent. Prove your agent is real before anyone asks.
For teams running agents in production. Stripe Checkout for product pro. Keys are issued after payment. SLA/white-label/on-prem are not self-serve.
For orgs with 10+ agents or regulatory requirements beyond the standard frameworks.
Hash-chained receipts on every plan. Stripe bills Pro. No hidden fees. Cancel anytime. Need help choosing?
Launch Thread · @EddyWoodss
6 tweets. No filler. Live now on X.
1/6AI agents need alibis. ProofChain gives them one.
2/688% of orgs had an AI agent incident last year. 14% have audit trails. When your agent approves a $12K invoice at 3am and someone asks "who authorized this" — what do you show them? A screenshot? Good luck.
3/6Mint AgentID on X1 (0.1 AGI). Then POST an action for a public hash-chained receipt — that receipt is not an NFT and is not written to X1.
4/6Same trail, honest report: count, first/last, tip hash, chain_ok. Not a COMPLIANT PDF.
5/6Lisa logs tweets. Waveforge logs jobs. This page is not a page view.
6/6Live API. Hash-chained receipts. Docs at proofchain.us/docs/quickstart.md. Not open source today.
Connect your X1 wallet, mint a soulbound AgentID (you pay 0.1 AGI + XNT fees), then get a ProofChain receipt. Log actions. Read the trail. Pro is $99/mo via Stripe — keys on the success page.