v1.0 - Exercise Engine

Exercise Evaluation Engine

Independent microservice that processes, validates and evaluates English learning exercises. HTTP REST and TCP native transports with dynamic grid generation.

What is DISNOMIA?

An exercise evaluation engine designed to integrate into any ecosystem.

DISNOMIA is an independent microservice that processes, validates, and evaluates learning exercises. It works with both relational and non-relational databases for seamless integration, regardless of your framework or platform goals.

It exposes two interfaces on the same logic core: HTTP REST for any language or framework, and TCP native optimized for Node.js and NestJS environments. It receives content and configuration, generates dynamic representations, and evaluates responses in real time.

Backend (Supabase)
HTTP / TCP
Exercise Engine (NestJS)
ExerciseEngineService, Processors, Validators, Grid Engines
TTS Service (Python)
Google Cloud TTS, ElevenLabs

Features

Instant Evaluation

Validators for multiple choice, fill blank and open answer with 0-10 scoring. Levenshtein distance for fuzzy matching.

Dynamic Grids

Word Search and Crossword generation with backtracking algorithms. Every attempt is unique.

Text-to-Speech

Integration with Google Cloud TTS and ElevenLabs for listening exercises with A/B dialogue support.

API Key Auth

API key authentication with generation, listing and revocation via admin endpoints.

Dual Transport

HTTP REST for any framework plus TCP native for Node.js and NestJS microservice communication.

Modular and Clean

Architecture with processors, validators and engines separated by responsibility. Easy to extend.

Why Disnomia?

An evaluation engine designed to integrate into any ecosystem.

MCP Integration

Connect multiple MCP servers to integrate voices (TTS), images, and AI agents into your exercise workflows.

Dev-Friendly

Works with Laravel, Django, NestJS, or raw HTTP. TCP native for Node.js microservices.

Dynamic Grids

Word Search and Crossword engines with backtracking. Every grid is unique, generated on-demand.

Real-Time Scoring

Instant evaluation with Levenshtein fuzzy matching, 0-10 scoring, and detailed feedback per answer.

Secure by Default

API key authentication with admin endpoints for generation, listing, and revocation.

Stateless & Scalable

No database, no state. Pure evaluation engine — horizontally scalable behind any load balancer.

Ready to integrate?

Explore the full API documentation or test endpoints interactively with Swagger.