Free tier available · No telemetry
Mac app 0.2.7 available now

Your codebase
has a story.

prism0x2A reads it.

Three intelligence layers. One coherence score. Zero guesswork.

Early Access for engineers · Mac app 0.2.7 available. On Windows or Linux, the command-line scan: npx -p prism0x2a prism scan .

Works with:TypeScriptJavaScriptReactNext.jsNode.js
Architecture

Three layers of intelligence

Each layer builds on the last. Together they give you a complete picture of your architecture.

B
BLUE — Blueprint Intelligence
What your code does

Deep AST traversal that builds a complete semantic model of your codebase — functions, dependencies, interfaces, and call graphs.

  • ◆Full AST scan of your TypeScript / React / Next.js / Node stack
  • ◆Import cycle and boundary detection
  • ◆Module dependency graph construction
  • ◆Interface and contract extraction
  • ◆Ask BLUE — chat about your codebase
A
AMBER — Capability Layer
How it's organized

Maps your code into named capabilities — discrete units of business logic — and detects when boundaries are violated.

  • ◆17 capability categories, auto-detected
  • ◆Boundary drift alerts per commit
  • ◆Cross-capability dependency scoring
  • ◆Conway's Law team alignment score
  • ◆Ask AMBER — interview your capability registry
G
GREEN — Coherence Score
How well it holds together

Synthesizes BLUE and AMBER signals across 14 architectural frameworks into a single A–F coherence grade.

  • ◆Single letter grade per repository
  • ◆Score trend and regression detection
  • ◆Per-framework breakdown and drill-down
  • ◆PR-level score delta preview
  • ◆Ask GREEN — chat with your architecture
BLUE→AMBER→GREENsignals compound up the stack
Your own AI

Run it with the AI you already have.

The scan needs no AI at all. AMBER's two AI steps, mapping your code to capabilities and checking claims against their proofs, can run as a prompt: copy it, run it, paste the answer back. No API key.

A Claude plan

Already paying for Claude? Run the prompt in Claude Code, on the plan you have. No API key and no second bill.

Your company's setup

Only approved models allowed, such as Claude Code on Amazon Bedrock? The prompt goes only where you paste it. prism0x2A itself calls no model here.

Another coding agent

The assistant has to be able to read your code, so a plain chat window will not do. A coding agent with access to your repository can answer it, and the answer goes through the same checks. How good it is depends on the model.

Ask in BLUE, AMBER and GREEN, Explain and Refactor are conversations with the model and need an API key.

One recorded scan

See it in action

The dashboard after scanning pixel-peloton, one of our own repositories: 1,429 files, 51 capabilities, 2,446 statements. Recorded on 5 Nov 2026, not a live run.

prism0x2A — localhost:3001
recorded
Scanning repository…

Or read a real scan, with the line behind every number:

Open the scan excerpt →

One file through BLUE, AMBER and GREEN · every figure cited · the OSS sandbox is one click further in

Capabilities

Everything your architecture needs

A tightly integrated suite of features, built on a single data model.

14 Framework Analyzers
SOLID, Clean Arch, Hexagonal, C4, 12-Factor, DORA-Style Predicted Metrics, Conway's Law, Wardley, ISO 25010 (6 of 8 characteristics), EIP, DDD, EDA, Bazel/Turborepo/Nx Monorepo + Auto-Detection
Architecture Knowledge Graph
Visual dependency & impact explorer across your entire codebase
Counterfactual Reasoning
"What if" analysis on architectural decisions
Cross-Codebase Fleet Pattern Recognition
Spot patterns and drift across multiple repos in your organisation
Predictions with Causal Reasoning
AI-driven architecture forecasts grounded in your code history
Aha Moment Finder
Surfaces hidden insights buried in your codebase
Extended Thinking
Claude extended reasoning mode for deep architecture queries
Cognitive Load
Per-capability complexity analysis to surface structural bottlenecks
Live Architecture Map
Miro sync + SVG export for Figma, Notion and docs
Architecture OKRs & Goal Setting
Set measurable architecture goals and track them over time
Architecture Interview Mode
Claude interviews you to capture architecture knowledge into a living capability map
Streak & Achievements
Gamified architecture health tracking
REST API v1 + API Keys
Full programmatic access to all prism0x2A data
Blueprint Scanner
Deep-reads your codebase in 30 seconds
AMBER Capabilities
17 capability categories, auto-detected
Coherence Score
A-F grade for your entire architecture
Pricing

Simple, transparent pricing

Start free. Upgrade when your codebase grows. Cancel anytime.

Free
€0

Everything you need to get started with architecture intelligence.

  • ✓1 repository
  • ✓Up to 15,000 lines of code
  • ✓AMBER capabilities — basic
  • ✓GREEN coherence score
Download for Mac
Most Popular
Pro
€19/mo

or €190 a year

For engineers who take architecture seriously.

  • ✓1 developer activation
  • ✓Blueprint Explorer at any repo size
  • ✓All 14 framework analyses
  • ✓Blueprint Explorer (6 views)
  • ✓Architect Mode (Ask / Explain / Refactor)
  • ✓Monorepo + aggregate workspace scan
  • ✓Cross-layer context + auto-generated insights
  • ✓Ask GREEN — Architecture Chat (Extended Thinking)
  • ✓Local CI gate engine + ready-made Vercel / GitHub Actions snippets
Get Pro
Built to fit the tools you already use
GH
GitHub
FG
Figma
MR
Miro
VC
Vercel
AZ
Azure DevOps
FAQ

Frequently asked questions.

prism0x2A · The answer to your codebase.

Your architecture deserves
better than gut feeling.

Free to start. No telemetry, no lock-in.

✓ Local-first✓ No data leaves your machine✓ Open scoring method