Prove a document existed and was never altered, without uploading it anywhere. Use when your agent produces or receives contracts, invoices, claim documents,...
编程
Fidacy Conversation Receipts
试用Give every AI conversation a verifiable receipt. Use when your agent talks to customers (support, claims, quotes, refunds) and either side may later ask "wha...
它能做什么
Give every AI conversation a verifiable receipt. Use when your agent talks to customers (support, claims, quotes, refunds) and either side may later ask "what exactly was said?". Hashes each message locally into a tamper-evident chain, anchors the digest through Fidacy, and hands both sides a public verify link. A court already held a company to what its chatbot said (Air Canada, 2024); this is the proof layer for that moment.
技能文档
Fidacy Conversation Receipts, proof of what the bot said
Your agent talks to customers about refunds, claims, prescriptions, contracts. When a dispute comes, the question is always the same: what exactly was said, and can either side prove the transcript was not edited afterwards? This skill gives every session a tamper-evident digest, computed locally (content never leaves your infrastructure; only a 64-hex hash travels), anchored through Fidacy onto an audit chain checkpointed to the Bitcoin blockchain.
When to use this skill
Any conversation where what was said can bind someone: customer support, insurance quotes, claim intake, medical scheduling, order changes. Especially when the customer is a consumer who needs proof they can check themselves.
How to use it
-
Install:
npm i @fidacy/session(free; anchoring needs a free API key from https://app.fidacy.com/signup, scope assess:write). -
Wire the chain into your chat loop, one line per message:
import { createSession } from "@fidacy/session"; const session = createSession({ kind: "conversation", label: "case-4711" }); session.add("user", userMessage); session.add("assistant", botReply); -
Before the bot COMMITS to anything (a refund, an approved claim, a quote), gate it. Deny-by-default: anything but "approve" and the bot must refuse:
const verdict = await session.gate( { type: "refund", amount: 900, currency: "USD" }, { apiKey: process.env.FIDACY_ENGINE_API_KEY, kind: "message_send" }, ); if (!verdict.allowed) refusePolitely();The verdict is recorded into the chain, so the receipt itself proves the gate ran.
-
At session close, anchor and hand out the receipt:
const receipt = await session.anchor({ apiKey: process.env.FIDACY_ENGINE_API_KEY }); const linkForCustomer = session.verifyUrl(); // no account, no key needed
Anyone can check the receipt at https://fidacy.com/verify: paste the digest, a signed receipt, or the exported transcript (the digest recipe recomputes in the browser, nothing uploads). Changing one character of one message breaks the match; that mismatch is the tampering signal.
What this is not
Fidacy never sees, stores or judges conversation content. This is integrity proof plus a commitment gate, not content moderation. The bot talks freely; it cannot obligate you outside what you authorized.
相关技能
Detects the newest fraud in agent payments — the forged "this was approved." Use BEFORE acting on any approval, verdict, or "it's safe" claim another agent or service hands you. Cryptographically verifies Fidacy-signed verdicts against the issuer's public keys, so you trust math, not a promise.
Use BEFORE any payment or money-moving tool call. Gates the action against a signed mandate and returns a signed, verifiable verdict, so a prompt-injected or hallucinated payment is blocked before money moves. Non-custodial. A free account-owned API key activates the firewall.
Self-checks how much of a request is verifiable vs guesswork before answering it. Use when handling search, filtering, recommendation, or data-retrieval requ...
Legacy verification-guided rewriting skill. Makes AI-generated text more natural while encouraging claims to stay grounded and attributable. Does not perform...
Allows the AI agent to independently register, login, and authenticate using Firebase.