RAPTOR — Architecture Designer

Browser-based tool for designing microservice architectures, simulating traffic, and generating documentation. No build step, no npm.

🇬🇧 English · 🇷🇺 Русский · 🇺🇦 Українська · 🇵🇱 Polski · 🇪🇸 Español · 🇩🇪 Deutsch

---

Quick Start

Without AI (diagram + simulation only)

npx serve . -p 5500

Or open the Live Demo → directly in your browser (no AI, no install).

With AI (RAPTOR AI via GitHub Copilot)

Step 1. Make sure you have GitHub Copilot — free plan works.

Step 2. Install Node.js if you haven't already.

Step 3. Install and run anti-api — it handles GitHub OAuth for you:

npm install -g anti-api
anti-api

A browser window opens → sign in to GitHub → anti-api starts on localhost:8964.

Step 4. Run RAPTOR:

npx serve . -p 5500

Open http://localhost:5500 → click 🤖 AI — ready to go.

---

Features

Diagram

• Blocks: clients, Gateway, services, BFF, Auth, PostgreSQL, Redis, MongoDB, Kafka, RabbitMQ, NATS, Queue, Elasticsearch, S3, CDN, LB, External
• Drag & drop, collision detection, bidirectional arrows
• Auto-layout by layers — Auto button
• C1 / C2 / C3 / Seq levels in toolbar

Scenarios ("Lines")

• Routes of requests through the system
• Each scenario: path of blocks, volume (rps), color, steps with verb + uri + resp
• Autofill builds steps from the diagram based on Gateway auth_mode

Simulation

• Balls travel along scenario paths, color = scenario color
• Kafka/RabbitMQ/NATS/Queue receive traffic when a ball passes through a producer service
• Consumption: periodic tick every ~500 ms independent of ball physics
• Click a ball → hop trace details with latencies
• Chaos mode: node failures, degradation, cache flush, overload

Broker Inspector

• Click Kafka/RabbitMQ/NATS/Queue during simulation → live panel
• SVG topology: producers → partitions → consumers
• Live counters: published / consumed / lag / DLQ, 500ms update
• Consumer groups with per-partition offsets, Replay button
• Business analyst justification: partitions, RF, delivery guarantees
• Broker settings without stopping simulation

UML Sequence

• Generated from scenario: business meaning above arrow, verb uri below //
• PlantUML standard: autonumber, ++/-- activations, group alt/else, ...pause...
• Kafka → dashed arrow -) without synchronous response
• "📋 PlantUML" button — copies for Confluence
• ref over blocks with horizontal drag (changes participants) and vertical (order)

Database

• PostgreSQL schema editor per service
• Visual ERD in crow's foot notation
• SQL-DDL import
• 23 ready-made templates (users, sessions, bookings, products, etc.)
• FK with REFERENCES + ON DELETE CASCADE, UUID, created_at

Auth Service

• Types: jwt_only / oauth2_jwt / session_based / api_key / oidc_sso
• Type changes the DB schema and Auth load in simulation
• Gateway reads auth_mode → automatically builds hop to Auth or skips

Templates

Services (9): search, media, chat_ws, cqrs, b2b_api, sso, scheduler, reports, saga_step
Groups (10): shop, fullmsa, eventdriven, mentorship, cqrs, b2b_api, sso, multibff, fintech, saga

Word Export

• MS-1 structure: description, C4 diagram, API per service, DB tables
• Offline via built-in docx.min.js

RAPTOR AI

🤖 AI button → chat panel on the right. Builds diagrams, writes UML, creates DB schemas — in natural language.

13 tools: add/remove block, add connection, create scenario, auto-layout, edit UML Sequence, change DB schema, clear canvas.

AI works via anti-api — an open-source proxy that handles GitHub Copilot OAuth for you. Thanks to ink1ing for the tool: without it you'd need to manually obtain a gho_ token via OAuth flow. Each user authenticates with their own account, no shared tokens.

---

State

Everything in the global S object (blocks, connections, scenarios, meta).
Save → JSON file. Load → drag JSON onto the canvas.

---

Releases

Version	What's new
v1.1.0	RAPTOR AI (13 tools), Kafka simulation fix, sim-trace
v1.0.0	First public release — diagram, simulation, UML, ERD, Word export

Full changelog → CHANGELOG.md