Logisketch – Smart Diagram Generator from Natural Language & Images

About

Generate system architecture, ER, flowchart, network packet, circuit logic, 3D, math, and sequence diagrams from natural language prompts.
A backend-powered diagram assistant for developers, educators, and system designers.
Supports 13+ diagram engines like PlantUML, Mermaid, Graphviz, Structurizr, D2, Vega-Lite, Tikz, and many more.

---

Supported Models

Text-based LLMs:

• meta-llama/llama-4-scout-17b-16e-instruct
• llama-3.3-70b-versatile
• llama-3.1-8b-instant
• meta-llama/llama-4-maverick-17b-128e-instruct

Text + Image-based LLMs:

• mistralai/mistral-small-3.2-24b-instruct-2506:free

---

Supported Diagram Engines (via Kroki)

Engine	Can Generate
Graphviz	Flowcharts, directed/undirected graphs
Blockdiag	Block diagrams
Seqdiag	Sequence diagrams
Packetdiag	Network packet diagrams
C4	C4 model diagrams (software architecture)
D2	Entity relationships, org charts, architectures
ERD	Entity Relationship Diagrams
Mermaid	Flowcharts, Gantt, class, pie, state diagrams
Nomnoml	UML class diagrams
PlantUML	UML diagrams (sequence, activity, use-case, etc.)
Structurizr	Software architecture diagrams (C4 model)
Symbolator	Digital logic and circuit diagrams
TikZ	Math figures, circuit diagrams, flowcharts
Vega-Lite	Bar, line, scatter charts (data visualization)
Wavedrom	Digital timing diagrams

---

Key Features

• Natural language to diagram via Kroki
• Image + text-based diagram generation
• Multi-model LLM support
• Custom prompt engineering support
• Rate limiting and bot protection via Arcjet
• Supports 13+ diagram engines

---

API Routes Description

1. POST /api/generate-image

Generates a diagram image from a natural language prompt, with optional image input (used for models that support images).
This route performs the full LLM-to-Kroki pipeline and returns both the generated code and the rendered diagram.

2. POST /api/render-code

Renders a given diagram code (in a supported diagram syntax) directly into an image using the selected Kroki engine.
This route bypasses the LLM and is ideal when you already have the code and just want to convert it to an image.

---

API Usage

For Text-based Models

POST /api/generate-image

{
  "prompt": {
    "text": "Create a CI/CD pipeline diagram with key stages and tool relationships."
  },
  "model": "llama-3.3-70b-versatile",
  "engine": "plantuml"
}

✅ Success Response

{
  "success": "true",
  "code": "...",
  "svg": "..."
}

❌ Error Response

{
  "success": "false",
  "error": "..."
}

---

For Text + Image Models

POST /api/generate-image

{
  "prompt": {
    "text": "Create an ER diagram from following image",
    "image": "https://dataedo.com/asset/img/blog/erd_with_pen.png"
  },
  "model": "llama-3.3-70b-versatile",
  "engine": "erd"
}

✅ Success Response

{
  "success": "true",
  "code": "...",
  "svg": "..."
}

❌ Error Response

{
  "success": "false",
  "error": "..."
}

---

For Rendering Code

POST /api/render-code

{
  "code":"..."
  "engine": "plantuml"
}

✅ Success Response

{
  "success": "true",
  "svg": "..."
}

❌ Error Response

{
  "success": "false",
  "error": "..."
}

---

Accepted Parameters

Parameter	Type	Allowed Values	Description
model	string (LLM)	Text-based LLMs meta-llama/llama-4-scout-17b-16e-instruct, llama-3.3-70b-versatile, llama-3.1-8b-instant, meta-llama/llama-4-maverick-17b-128e-instruct Text+Image LLMs mistralai/mistral-small-3.2-24b-instruct-2506:free	LLMs used for generating diagrams
engine	string (Diagram Engine)	plantuml, mermaid, graphviz, structurizr, blockdiag, seqdiag, packetdiag, c4plantuml, d2, erd, nomnoml, tikz, vegalite, symbolator, wavedrom	Rendering engines supported by Kroki

---

Environment Configuration

Create a file named either:

• .env.development.local – for development
• .env.production.local – for production

⚠️ Other filenames will not work without changes to the environment loader config.

Place it in the root of your project directory.

---

Sample .env.*.local File

# ENVIRONMENT
NODE_ENV=development

# PORT
PORT=3000

# ARCJET
ARCJET_KEY=your_arcjet_key_here

# LLM PROVIDERS

# 1. GROQ
GROQ_API=your_groq_api_key_here
GROQ_ENDPOINT=https://api.groq.com/openai/v1/chat/completions

  # GROQ MODELS
  GROQ_LLM_1=meta-llama/llama-4-scout-17b-16e-instruct
  GROQ_LLM_2=llama-3.3-70b-versatile
  GROQ_LLM_3=llama-3.1-8b-instant
  GROQ_LLM_4=meta-llama/llama-4-maverick-17b-128e-instruct

# 2. OPENROUTER

OPENROUTER_ENDPOINT=https://openrouter.ai/api/v1/chat/completions

  # OPENROUTER MODELS
  OPENROUTER_LLM_1=mistralai/mistral-small-3.2-24b-instruct-2506:free
  OPENROUTER_API_1=your_mistralai_api_key_here

# RENDER ENGINES
ENGINE_1=graphviz
ENGINE_2=blockdiag
ENGINE_3=seqdiag
ENGINE_4=packetdiag
ENGINE_5=c4plantuml
ENGINE_6=d2
ENGINE_7=erd
ENGINE_8=mermaid
ENGINE_9=nomnoml
ENGINE_10=plantuml
ENGINE_11=structurizr
ENGINE_12=symbolator
ENGINE_13=tikz
ENGINE_14=vegalite
ENGINE_15=wavedrom

---

Get Your API Keys

• ARCJET API Key → https://arcjet.com

  Sign up at https://arcjet.com and create a new project. Once created, your API key will be displayed at the top under your project name as ARCJET_KEY = **************

• GROQ API Key → https://console.groq.com/keys

• Mistral API Key → https://openrouter.ai/mistralai/mistral-small-3.2-24b-instruct:free/api

---

How to Run Locally

1. Clone the repo and cd into it

git clone git@github.com:anubhav-m/logisketch-server.git
cd logisketch-server

2. Install dependencies

npm install

3. Create and configure your .env file

Refer to the Environment Configuration section for setup

4. Run the development server

npm run dev

---

Deployment Guide

When deploying on Render, Railway, Fly.io, or similar platforms, make sure to configure the following:

Build Command:

npm install

Start Command:

npm start

Environment Variable:

NODE_ENV=production

Secrets:

Upload your env.production.local file.

Refer to the Environment Configuration section for the full list of variables.

---

License

This project is licensed under the terms of the MIT License.