Aucert internal documentation
This site contains internal engineering documentation for the Aucert platform. It covers architecture deep-dives, infrastructure details, operational runbooks, and team workflows.
warning
This documentation is internal only. It is restricted to @aucert.ai team members via Cloudflare Access. Do not share content from this site externally.
Getting started
New to the team? Start with the first day checklist — it walks you through account setup, tool installation, and your first build verification.
Key sections
| Section | What you'll find | Start here |
|---|---|---|
| Onboarding | First-day setup, dev environment, team directory | First day checklist |
| Architecture | Full pipeline internals, KG engine, Device Twin, Verification Cascade, model orchestration | 5-layer deep dive |
| Infrastructure | Azure resources, Terraform tiers, networking, Cloudflare, Foundry | Azure topology |
| Agents | AI workforce registry (Astra), dispatch architecture, adding agents | Workforce overview |
| How-to guides | Step-by-step runbooks for building, testing, deploying, and debugging | Local dev setup |
| Decisions | Architecture Decision Records (ADRs) — why we chose what we chose | ADR index |
| Workflows | Spec-driven development, drift capture | Spec-driven development |
| Security | Secrets management, Key Vault, access control | Secrets management |
How to use this documentation
- Context files first: The
.context/directory in the repo is the source of truth for current state. These docs explain the why and how behind that state. - L1 → L2 → L3: Context loads in layers. Start with repo-wide L1 files (
.context/ARCHITECTURE.md), then platform-specific L2 (backend/.context/BACKEND.md), then module-specific L3 specs. - Runbooks are runnable: Every how-to guide has copy-paste commands with verification steps. If a command doesn't work, the doc is stale — update it.
- Search: Use the sidebar or
Ctrl+Kto search. Pages are tagged by topic.