# Hyxnj666 Creator AI Memory AI Memory > System Architecture - Skill: `tomevault-io/hyxnj666-creator-ai-memory-ai-memory` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/hyxnj666-creator-ai-memory-ai-memory` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/hyxnj666-creator-ai-memory-ai-memory/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/hyxnj666-creator-ai-memory-ai-memory --- # System Architecture > Auto-generated by ai-memory from AI chat history. Edit memories via `ai-memory list` / `extract` / `resolve` rather than this file — it is fully regenerated on every `ai-memory rules --target skills`. ## Event sourcing for the billing audit log The billing domain becomes event-sourced. The source of truth is an append-only `billing_events` Postgres table (`InvoiceCreated`, `LineItemAdjusted`, `PaymentCaptured`, etc.), each row carrying a SHA-256 hash chain over `(prev_hash, payload, timestamp)`. The current invoices table becomes a read-side projection rebuilt from the log. The rest of the app stays on the existing CRUD model — only billing needs this. Why: Event sourcing answers compliance directly: every state change is a first-class, immutable record. Hash-chaining provides tamper evidence without an external service. Scoping to billing keeps the operational cost contained. Rejected: Shadow-table journaling rejected — couples write-side logic to bookkeeping and does not give tamper evidence. External audit-log service rejected — added vendor cost without measurable gain over an in-database hash chain. --- > Source: [hyxnj666-creator/ai-memory](https://github.com/hyxnj666-creator/ai-memory) — distributed by [TomeVault](https://tomevault.io).