LCM — Verifiable Decision Layer for AI Systems

AI decisions are invisible. LCM makes them provable.

Deterministic, replayable, and cryptographically sealed proof for every policy-bound AI decision.

Install

npm install lcm-audit

Quick Start

import { LCM } from "lcm-audit";

LCM.init({ apiKey: "your-api-key" });

const input = { action: "refund", orderId: "ORD-938472", amount: 129.00, currency: "USD", reason: "defective_item" };

const decision = LCM.decide(input);

const sealed = await LCM.seal(decision);

console.log("\n🔍 Open Proof:"); console.log(sealed.traceUrl); console.log(sealed.derivation_root);

const verified = LCM.verify(sealed); console.log(verified.valid);

const replayed = await LCM.replay(sealed.audit_id);

Get Pro Access

Restore trust in AI decisions.

Unlock full audit history and verifiable proof.

https://www.lcm3.com/apply.html

Why LCM

AI systems can act.

But when they approve a refund, cancel an order, authorize a payment, or make a decision:

Can you prove exactly why it happened?

If not:

• accountability becomes unclear
• audits become difficult
• disputes become expensive
• trust becomes fragile

LCM solves this. It turns decisions into proof artifacts.

How It Works

Decision → Seal → Verify → Replay → Detect

Seal → cryptographic commitment Verify → integrity check Replay → decision reconstruction Detect → tamper detection

Example

Original:

{ "action": "refund", "amount": 129.00 }

Modified:

{ "action": "refund", "amount": 1290.00 }

Result:

Tampered Replay Failed Chain Invalid

See it fail in real-time:

https://www.lcm3.com/explorer.html

Try LCM in 30 Seconds

See what happens when an AI decision is tampered.

https://www.lcm3.com/sandbox.html

No setup required.

Inspect a Real Decision

View a full decision proof, chain integrity, and audit trace:

https://www.lcm3.com/explorer.html

Support

https://github.com/paibyun9/LCM-a2a-spec/issues