Production Platform

The Platform

A substantial engineering investment in enterprise-grade AI safety infrastructure

AI Assess Tech isn't a prototype or proof-of-concept. It's a production platform with 814,776 lines of original code, including production and tests, built with 750+ VibeCoded hours of engineering effort. At 100 hours Before AI (BAI) for every 1 hour of VibeCoding, the total BAI level of effort is about 35–40 man years. Backed by 17,112 automated tests, 254 AI governance rules, 659K VibeCoding session lines across 152 development threads, and 2,779,462 lines of supporting documentation.

This page provides transparency into the engineering foundation that makes AI Assess Tech a trustworthy platform for AI behavioral assessment.

Codebase at a Glance

Real numbers from a real production system. Updated 9/9/2026. Code, tests, and documentation are a first-party source count. VibeCoding session lines are aggregated from local Cursor development history (not shipped code).

814,776

Lines of Code

480,282 production + 334,494 tests

17,112

Automated Tests

4,968 suites, source count

2,779,462

Supporting Documentation

Git-tracked .md and .mdc files

659K

VibeCoding Session Lines

152 Cursor threads, local history

219

API Endpoints

REST API route files

254

AI Governance Rules

243 live, 11 superseded

100

Database Models

3,914 lines schema

Platform Engineering

Key investments that make the AI Assess Tech SaaS platform enterprise-ready.

4-Dimensional Ethical Assessment

Mathematically rigorous LCSH framework scoring AI behavior across Lying, Cheating, Stealing, and Harm with personality classification and variance analysis.

Enterprise Multi-Tenancy

100-model database with full organization isolation, hierarchical RBAC, and per-route permission enforcement across 219 API route files.

Compliance Automation

Risk-based scheduling with iCal export, fleet health dashboards, audit export, and per-dimension threshold enforcement.

Blockchain-Anchored Integrity

Question banks locked with SHA-256 hashes and anchored to Ethereum via smart contract — publicly verifiable, tamper-evident.

Multi-Provider AI Testing

Automated 120-question assessments against Anthropic, OpenAI, Gemini, and Grok with real-time SSE progress streaming.

Verifiable Attestation

HMAC-SHA256 signed certificates with public verification API, QR codes, and embeddable trust badges.

ANET-AGI-001

Multi-Agent Fleet

ANET-AGI-001 is a production deployment of 6 autonomous AI agents — running in production since February 16, 2026 — each with independent ethical oversight, temporal behavioral tracking, and constitutional governance. It represents a second major engineering investment equal in scale to the core platform, and the first demonstrated implementation of runtime AI governance with separation of powers.

Noah

Temporal Ethical Guidance

Patent-pending Three-Variable Guidance Equation with flight plans, corridor tracking, and inertial ethical monitoring

Files

532

Tests

15,900

Lines

Grillo

Assessment Coordinator

LCSH assessment framework with drift detection, fleet anomaly detection, and SHA-256 audit chains

Files

408

Tests

17,392

Lines

Jessie

Fleet Commander

7-step morning briefing protocol with veto authority, DecisionStore, and 8 fleet tools

Files

316

Tests

4,283

Lines

Nole

Trust Evangelist

Coinbase CDP wallet on Base L2, multi-platform social, and 25 agent tools with Commander approval gates

Files

408

Tests

13,684

Lines

Sam

Chief Engineer

Engineering pipeline manager with Docker sandbox, memory persistence, and 10 fleet tools

Files

146

Tests

5,714

Lines

Mighty Mark

System Health Sentinel

64+ health checks across 12 categories, morning report via Telegram, watchdog auto-restart, and predictive preflight probes

Files

1420

Tests

26,845

Lines

Fleet Infrastructure

Shared infrastructure supporting all 6 agents

Package	Description	Files	Tests	Lines
Fleet Bus	Constitutional inter-agent messaging	38	390	8,145
Fleet Alerting	Severity-routed escalation system	13	7	1,225
Prompt Shield	11-layer injection defense	81	143	5,022
Mission Control	28-panel glass cockpit dashboard	272	757	57,908

Fleet Engineering Highlights

Constitutional Communication

Typed, hash-chained inter-agent messaging with patent-aligned routing constraints and tamper-evident audit trails.

Temporal Ethical Navigation

Patent-pending guidance equation tracking AI ethical trajectories over time with flight plans, drift detection, and inertial monitoring.

11-Layer Prompt Defense

Fleet-wide injection protection including MCP tool poisoning detection, encoding attack unwrapping, and canary leak detection.

Unified Fleet Alerting

Sub-second severity-routed alerts with time-based escalation, deduplication, and Inspector General aggregation.

Mission Control Cockpit

28-panel real-time dashboard with zero-dependency SQLite architecture, Ed25519 device identity, and WebSocket live feeds.

Constitutional Governance

Three-tier authority model with veto power, morning briefings, proposal gates, and separation of powers.

Testing Infrastructure

Every feature is backed by comprehensive automated testing across both the platform and fleet.

AI Assess Tech Test Harness Dashboard snapshot from 9/9/2026. Current source counts are in the table below.

Harness snapshot from 9/9/2026. Current counts are the source inventory dated 9/9/2026.

Test Suite Breakdown

17,112 tests across 4,968 suites, counted from source on 9/9/2026. This is not a harness run.

Category	Location	Files	Lines
Platform tests	__tests__	1,028	224,353
Package tests	packages/	323	66,838
App tests	apps/	153	17,223
Deployment tests	deployment/	41	15,721
Other first-party tests	Remaining first-party test files	49	10,359
TOTAL	17,112 tests, 4,968 suites	1,594	334,494

Technology Stack

Built on modern, well-supported technologies for long-term maintainability.

Frontend

Next.jsReact framework with App Router

16.x

ReactUI library

19.x

TypeScriptType safety

5.x

Tailwind CSSStyling

4.x

shadcn/uiComponent library

Latest

Backend

Next.js API RoutesRESTful API

16.x

PrismaORM and database toolkit

6.x

NextAuthAuthentication

5.x

ZodRuntime validation

4.x

Infrastructure

VercelHosting and deployment

—

NeonServerless PostgreSQL

—

StripePayment processing

—

Testing

JestUnit, integration, API tests

29.x

PlaywrightBrowser E2E tests

1.49.x

VitestFleet & SDK tests

Latest

Testing LibraryReact component testing

16.x

ANET-AGI-001 FleetNew

OpenClawAI agent gateway

2026.2.25

SQLite (WAL)Agent-local persistence

Latest

ZustandMission Control state

5.x

RechartsData visualization

3.x

viemEthereum/Base L2

Latest

SDK Ecosystem

Official SDKs for seamless integration into your applications.

TypeScript SDK

Production

@aiassesstech/sdk

Full-featured SDK for Node.js and browser environments

25 tests passing

Python SDK

Production

aiassesstech

Native Python SDK for data science and backend integrations

100 tests passing

SDK Wrapper

Production

@aiassesstech/sdk-wrapper

Lightweight wrapper for quick integrations

7 tests passing

Our Commitment to Transparency

We believe the platforms you trust should be transparent about their engineering investment. The metrics on this page are real, verifiable, and updated regularly. When you use AI Assess Tech, you're using a platform built with the same rigor we expect from the AI systems we assess.

View Patent Portfolio →View Academic References →View Methodology →

Metrics last updated: 9/9/2026