Skip to the content
2 layers60 × 40 mmTest pcb 001, the printed circuit board in 3D

A full-stack, AI-powered parametric CAD generation platform. Type "iPhone 15 Pro case with camera cutout and USB-C port", and within seconds a fully-parametric, 3D-printable model appears in your browser — adjustable…

Size
60 × 40 mm
Copper layers
2
Thickness
1.6 mm
Parts
6
Nets
2
Pads
44
Tracks, vias
1, 0
Pours
1
Routed
Yes
KiCad
KiCad 7 (file 20221018)
Created
19 Mar 2026
Updated
1 Jul 2026

Overview

Test pcb 001 is an open-source printed circuit board by Aslanmatejka, from the Aslanmatejka/inventa.ai repository on GitHub. It is a 2-layer board of 60 × 40 mm with 6 parts on 2 nets, routed with 1 track and 0 vias.

Its parts include 1 IC and 1 diode.

README

README.md

🧱 CAD AI Builder — Chat-to-CAD Platform

Describe any physical product in plain English → AI generates production-quality 3D geometry → download STL/STEP instantly.

A full-stack, AI-powered parametric CAD generation platform. Type "iPhone 15 Pro case with camera cutout and USB-C port", and within seconds a fully-parametric, 3D-printable model appears in your browser — adjustable with sliders, exportable in STEP and STL, and editable with further natural language commands. Upload your own CAD files (STEP, IGES, STL, OBJ, 3MF, DXF…) and continue editing them with language.


Table of Contents

  1. Features
  2. Architecture Overview
  3. Tech Stack
  4. Project Structure
  5. Prerequisites
  6. Installation & Setup
  7. Running the App
  8. Environment Variables
  9. API Reference
  10. CAD File Upload & NLP Editing
  11. How the AI Pipeline Works
  12. Self-Healing Engine
  13. Parametric System
  14. Product Library
  15. Optional Services
  16. Frontend Architecture
  17. State Management
  18. Testing
  19. Roadmap
  20. Contributing
  21. License

Features

Core

FeatureDescription
🗣️ Natural Language → 3D ModelDescribe any product; Claude AI writes CadQuery Python that builds real solid geometry
🔁 Iterative NLP EditingSend follow-up messages to modify existing designs ("add ventilation slots", "make it 10mm taller")
📁 CAD File UploadUpload STEP, IGES, STL, OBJ, 3MF, DXF, BRep, PLY, GLB and edit them with language
🎛️ Live Parameter SlidersEvery dimension is a named parameter; sliders rebuild the model in seconds (no AI call)
💾 Dual ExportDownload STL for 3D printing, STEP for CAD editing in FreeCAD/Fusion 360/SolidWorks
🐍 Parametric Python ScriptEach build also exports a standalone .py CadQuery script you can run and modify locally

AI & Generation

FeatureDescription
🧠 Claude Sonnet / OpusUses claude-opus-4-20250514 by default; configurable via env var
🔥 High-complexity mode98+ product templates with real-world dimensions; 3 detail tiers (standard / detailed / professional)
🧪 Completeness analysisAfter generation, AI self-critiques its code and re-enhances if features are missing
🛡️ Infinite self-healing loopCadQuery errors trigger automatic fix retries with 5 graduated strategies (targeted → nuclear)
⚡ Prompt cachingIdentical prompts return instantly from an in-memory SHA-256 cache (no Al call)
📡 SSE streaming6-step progress events streamed live to the UI as the model is built

UI/UX

FeatureDescription
↔️ Independent panelsChat and 3D preview are fully decoupled — collapse/expand either, drag resize handle
📱 Mobile tab layoutOn screens < 968 px, chat and preview switch via tabs
↩️ Undo / Redo20-step history with Ctrl+Z / Ctrl+Shift+Z
📊 Build progress barAnimated progress bar with elapsed-time counter during generation
🔽 Smart auto-scrollChat only auto-scrolls if you're already at the bottom; "↓ New messages" button otherwise
🌐 3D WebGL viewerThree.js/React Three Fiber with orbit controls, lighting, and auto-fit
🎨 Empty-state & loading animationsPlaceholder canvas with floating icon and 3D rotating cube during build

Architecture Overview

text
┌─────────────────────────────────────────────────────────────────┐
│                        Browser (React 18)                       │
│  ┌──────────────┐  drag-resize  ┌────────────────────────────┐  │
│  │  Chat Panel  │◄─────────────►│      3D Preview Panel      │  │
│  │              │               │  ┌──────────────────────┐  │  │
│  │  • Messages  │               │  │  Three.js WebGL      │  │  │
│  │  • Steps log │               │  │  (STL via STLLoader) │  │  │
│  │  • Healing   │               │  └──────────────────────┘  │  │
│  │    history   │               │  ┌──────────┐ ┌─────────┐  │  │
│  │  • FileUpload│               │  │ Params   │ │ Export  │  │  │
│  │  • PromptBox │               │  │ Sliders  │ │ Panel   │  │  │
│  └──────────────┘               │  └──────────┘ └─────────┘  │  │
└─────────────────────────────────────────────────────────────────┘
              │  SSE stream / REST  │  REST  │
              ▼                    ▼        ▼
┌──────────────────────────────────────────────────────────────┐
│                  FastAPI  (port 3001)                        │
│                                                              │
│  POST /api/build/stream ──► claude_service                   │
│      │  (SSE 6-step)         generate_design_from_prompt()   │
│      │                          │                            │
│      │                          ▼                            │
│      │                    parametric_cad_service             │
│      │                    _execute_cadquery_code()           │

Tech Stack

Backend

LayerTech
Web frameworkFastAPI 0.128+
ASGI serverUvicorn
CAD geometryCadQuery 2.4+ (OpenCASCADE Technology / OCC)
Parametric partscq_warehouse (fasteners, bearings, gears)
AI / LLMAnthropic Claude (claude-opus-4-20250514) via anthropic SDK
Mesh importtrimesh (STL/OBJ/3MF/PLY/GLB)
IGES importOCP IGESControl_Reader
AsyncPython asyncio, httpx for streaming
Configpydantic-settings + .env
Database (optional)MySQL via SQLAlchemy + pymysql
Async tasks (optional)Celery + Redis
S3 sharing (optional)boto3

Frontend

LayerTech
UI frameworkReact 18.2
State managementuseReducer + React Context (AppContext)
3D rendering@react-three/fiber 8.15 · @react-three/drei 9.92 · three 0.159
SSE streamingNative fetch + ReadableStream
File uploadXMLHttpRequest (for upload progress)
Build toolingCreate React App 5 (Webpack 5)

Project Structure

text
Cad-ai-builder/
├── Backend/
│   ├── main.py                    # FastAPI app, all routes, in-memory stores
│   ├── start.py                   # Startup script (sets UTF-8 encoding)
│   ├── config.py                  # pydantic-settings config
│   ├── tasks.py                   # Celery task definitions (optional)
│   ├── validator.py               # Extra param validation helpers
│   ├── requirements.txt           # Core Python dependencies
│   ├── requirements-phase4.txt    # Optional deps (Celery, boto3, trimesh)
│   └── services/
│       ├── __init__.py            # Service singletons & availability flags
│       ├── claude_service.py      # AI design generation, self-healing, system prompts (~5100 lines)
│       ├── parametric_cad_service.py  # Code safety, 7-transform pipeline, exec sandbox (~1050 lines)
│       ├── cad_import_service.py  # CAD file upload & format conversion
│       ├── cadquery_service.py    # Legacy CadQuery geometry service
│       ├── database_service.py    # MySQL projects/builds/messages ORM
│       ├── glb_service.py         # STL→GLB conversion (trimesh)
│       ├── s3_service.py          # S3 upload/download sharing
│       ├── product_library.py     # 98+ product templates with real-world dimensions (~2400 lines)
│       └── product_visual_knowledge.py  # Per-category visual & build guides (~1700 lines)
│
├── client/
│   ├── package.json
│   └── src/

Prerequisites

RequirementMinimum versionNotes
Python3.10+3.12 recommended
Node.js18+22 recommended
npm8+
Anthropic API key—console.anthropic.com
MySQL8.0+Optional — app runs without it
Redis7+Optional — only needed for Celery async tasks

Windows users: CadQuery's OCC kernel is available as pre-built wheels. No FreeCAD or system OpenCASCADE installation required.


Installation & Setup

1. Clone the repository

bash
git clone https://github.com/Aslanmatejka/Cad-ai-builder.git
cd Cad-ai-builder

2. Create and activate a Python virtual environment

bash
# Windows (PowerShell)
python -m venv .venv
.venv\Scripts\Activate.ps1

# macOS / Linux
python3 -m venv .venv
source .venv/bin/activate

3. Install Python dependencies

bash
pip install -r Backend/requirements.txt

For optional features (GLB export, S3 sharing, Celery):

bash
pip install -r Backend/requirements-phase4.txt

4. Set up environment variables

bash
cp .env.example .env

Edit .env and add your Anthropic API key (the only required value):

ini
ANTHROPIC_API_KEY=sk-ant-api03-...

See Environment Variables for all options.

5. Install frontend dependencies

bash
cd client
npm install
cd ..

Running the App

Open two terminals from the project root:

Terminal 1 — Backend

powershell
# Windows (PowerShell), from project root with venv active
cd Backend
python start.py

The backend starts on http://localhost:3001 with hot-reload enabled.
Health check: GET http://localhost:3001/ → {"status": "healthy"}
OpenAPI docs: http://localhost:3001/docs

Terminal 2 — Frontend

powershell
cd client
npm start

The React dev server starts on http://localhost:3000 and opens automatically.

Windows note: If you see process already running errors, kill stale processes first:

powershell
Get-Process -Name python -ErrorAction SilentlyContinue | Stop-Process -Force
Get-Process -Name node   -ErrorAction SilentlyContinue | Stop-Process -Force

Environment Variables

Place these in the .env file at the project root. A copy at Backend/.env also works.

VariableRequiredDefaultDescription
ANTHROPIC_API_KEY✅—Anthropic API key for Claude
AI_MODEL_NAME❌claude-opus-4-20250514Claude model to use
AI_MAX_TOKENS❌16384Max tokens per AI response
AI_TEMPERATURE❌0.3AI temperature (0.0–1.0)
PORT❌3001Backend server port
HOST❌0.0.0.0Backend bind host
DEBUG❌falseEnable debug mode
DB_HOST❌localhostMySQL host
DB_PORT❌3306MySQL port
DB_USER❌rootMySQL user
DB_PASSWORD❌—MySQL password (leave blank to disable DB)
DB_NAME❌product_builderMySQL database name
AWS_ACCESS_KEY_ID❌—AWS key for S3 model sharing
AWS_SECRET_ACCESS_KEY❌—AWS secret for S3 model sharing
AWS_S3_BUCKET❌—S3 bucket name
REDIS_URL❌redis://localhost:6379Redis URL for Celery

Minimal .env for local development:

ini
ANTHROPIC_API_KEY=sk-ant-api03-your-key-here

API Reference

Build Endpoints

POST /api/build/stream

Stream a new CAD build via Server-Sent Events (SSE).

Request body:

json
{
  "prompt": "iPhone 15 Pro case with camera cutout and USB-C port",
  "previousDesign": null,
  "projectId": null
}

For modifications, pass previousDesign with the previous build's code, parameters, and explanation.

SSE Event stream (each line is data: {...}\n\n):

StepEvent
1Product library search
2Claude AI design generation
3Completeness check & enhancement
4CadQuery geometry execution (+ self-healing events)
5STEP / STL file export
6Complete — includes full result

Final event payload:

json
{
  "step": 6,
  "status": "complete",
  "result": {
    "buildId": "uuid",
    "stlUrl": "/exports/cad/{buildId}.stl",
    "stepUrl": "/exports/cad/{buildId}.step",
    "parametricScript": "/exports/cad/{buildId}_parametric.py",
    "parameters": [{"name": "width", "default": 75, "min": 10, "max": 200, "unit": "mm"}],
    "explanation": {"design_intent": "...", "features_created": "..."},
    "design": {"code": "import cadquery...", "parameters": [...]}
  }
}

POST /api/rebuild

Rebuild a model with updated parameter values — no AI call, executes saved _parametric.py script.

Request body:

json
{
  "buildId": "uuid",
  "parameters": {"width": 80, "height": 150}
}

POST /api/build (non-streaming)

Synchronous build — waits for full result before returning. Useful for scripting. Same request body as /api/build/stream.


File Upload Endpoints

GET /api/upload/formats

Returns all supported upload formats with editability flags.

json
{
  "solidFormats": [".step", ".stp", ".iges", ".igs", ".brep"],
  "meshFormats": [".stl", ".obj", ".3mf", ".ply", ".off", ".glb", ".gltf"],
  "2dFormats": [".dxf"]
}

POST /api/upload

Upload a CAD file (multipart/form-data, field name file, max 100 MB).

Response:

json
{
  "buildId": "uuid",
  "originalFilename": "bracket.step",
  "format": ".step",
  "stlFile": "/exports/cad/{buildId}.stl",
  "stepFile": "/exports/cad/{buildId}.step",
  "boundingBox": {"width": 50.0, "depth": 30.0, "height": 20.0},
  "geometryInfo": {"solids": 1, "faces": 24, "edges": 48, "vertices": 32, "volume": 12500.0},
  "editable": true,
  "importCode": "import cadquery as cq\n...",
  "success": true
}

POST /api/upload/edit

Apply a natural language edit to an uploaded CAD file.

Request body:

json
{
  "buildId": "uuid",
  "prompt": "Add four M3 mounting holes in the corners",
  "importCode": "import cadquery as cq\nresult = cq.importers.importStep(...)\n...",
  "previousDesign": null
}

Returns the same shape as the /api/build/stream final result.


Scene Management Endpoints

MethodPathDescription
POST/api/scene/createCreate a new multi-product scene
GET/api/scene/{id}Get scene + all products
POST/api/scene/{id}/add-productAdd a product to a scene
PUT/api/scene/product/{instanceId}/transformUpdate position/rotation/scale
POST/api/scene/product/{instanceId}/duplicateDuplicate with offset
DELETE/api/scene/product/{instanceId}Remove from scene
POST/api/scene/{id}/assembleGroup products into assembly
DELETE/api/scene/assembly/{id}Disassemble group

Project & History Endpoints (MySQL required)

MethodPathDescription
GET/api/projectsList all saved projects
POST/api/projectsCreate project
GET/api/projects/{id}Get project with builds + chat messages
PUT/api/projects/{id}Rename / update project
DELETE/api/projects/{id}Delete project and all data
POST/api/messagesSave a chat message to a project
GET/api/historyRecent builds across all projects

Utility Endpoints

MethodPathDescription
GET/Health check
GET/api/healthDetailed service status
GET/exports/cad/{filename}Serve generated CAD/STL/STEP files
GET/exports/uploads/{filename}Serve uploaded source files
GET/api/cache/check?prompt=...Check if a prompt has a cached result
POST/api/convert/glbConvert STL to GLB (trimesh, optional)
GET/api/mesh/stats/{buildId}Mesh vertex/face statistics
POST/api/s3/uploadUpload model to S3 for sharing
GET/api/s3/download/{key}Download shared model from S3

CAD File Upload & NLP Editing

One of the platform's most powerful features: upload any supported CAD file and immediately start editing it with natural language.

Supported Formats

FormatExtensionTypeNLP Editing
STEP.step, .stpSolid B-Rep✅ Full support
IGES.iges, .igsSolid B-Rep✅ Full support
BRep.brepOpenCASCADE native✅ Full support
DXF.dxf2D sketch✅ Auto-extruded to 3D
STL.stlMesh⚠️ View & export only
OBJ.objMesh⚠️ View & export only
3MF.3mfMesh⚠️ View & export only
PLY.plyMesh⚠️ View & export only
GLB/glTF.glb, .gltfMesh⚠️ View & export only

Tip: For full NLP editing of mesh files, convert them to STEP first using FreeCAD, Fusion 360, or any STEP exporter, then re-upload.

Editing Flow

  1. Drag-and-drop a STEP/IGES file onto the Upload CAD file zone in the chat panel
  2. The model appears in the 3D viewer with geometry metadata (bounding box, face/edge count, volume)
  3. A "NLP editable" badge appears in the prompt bar
  4. Type your edit: "add four M3 mounting holes 5mm from each corner"
  5. Claude reads the imported geometry code as the starting point and produces modified CadQuery code
  6. The new model replaces the old one in the viewer
  7. Chain further edits — each iteration uses the latest code as the base

How the AI Pipeline Works

Build Pipeline (6 steps)

text
User prompt
    │
    ├─ 1. Product Library Search
    │      Searches 98+ product templates for real-world reference dimensions
    │      (e.g., "iPhone 15 Pro" → exact port positions, camera island size)
    │
    ├─ 2. Claude AI Design Generation
    │      System prompt (~3000 lines) instructs Claude to return raw JSON:
    │      { parameters: [...], code: "import cadquery...", explanation: {...} }
    │      Adapts max_tokens (up to 16384) and temperature by complexity
    │
    ├─ 3. Completeness Check & Enhancement
    │      analyze_code_completeness() checks for missing features
    │      If incomplete → enhance_incomplete_design() sends back to AI for targeted fixes
    │
    ├─ 4. CadQuery Execution (with self-healing)
    │      7-transform preprocessing pipeline (see below)
    │      exec() in sandboxed namespace
    │      On error → infinite self-healing loop (5 escalating phases)
    │
    ├─ 5. File Export
    │      CadQuery → .step (editable) + .stl (printable) + _parametric.py (script)
    │
    └─ 6. Return result to frontend via SSE

7-Step Code Preprocessing Pipeline

Before exec(), every AI-generated code snippet goes through these transforms in order:

#TransformWhat it does
1_strip_centered_from_non_box()Removes centered= from .extrude(), .rect(), .circle() — only .box() supports it
2_ensure_box_grounding()Adds centered=(True,True,False) to the first main .box() so Z=0 is ground
3_fix_zero_dimensions()Replaces .extrude(0) with .extrude(0.1) to prevent zero-height geometry errors
4_fix_negative_z_positions()Warns and clamps features placed below Z=0
5_ensure_result_assignment()Auto-adds result = body if the AI forgot to assign the final workplane
6_clamp_fillet_radii()Injects min(r, _auto_fillet_max) guards — 15% of the smallest dimension
7_wrap_fillets_in_try_except()Wraps unprotected .fillet() / .chamfer() calls in try/except blocks

Self-Healing Engine

When CadQuery raises a RuntimeError or ValueError, the system never gives up. It classifies the error into one of 18 categories and retries with progressively more aggressive fixes:

Error Categories

GEOMETRY_FILLET_CHAMFER · GEOMETRY_SHELL · SKETCH_NOT_CLOSED · SELECTOR_FAILED · CURVE_TANGENT · REVOLVE_AXIS · MATH_ERROR · LOFT_FAILED · BOOLEAN_FAILED · WRONG_PARAMETER · SWEEP_FAILED · WORKPLANE_STACK · OCC_KERNEL_ERROR · ATTRIBUTE_ERROR · TYPE_ERROR · EMPTY_RESULT · NAME_ERROR · GENERAL

Healing Phases

PhaseAttemptsStrategy
Targeted1Fix only the crashing line; reduce fillet radii 30%
Conservative2–3Reduce radii 60%; wrap in try/except; simplify splines
Aggressive4–5Strip all fillets, chamfers, shells; replace lofts with extrusions
Rewrite6–7Rewrite the failing section using only safe primitives
Nuclear8+Rebuild the entire model from scratch with boxes & cylinders only

The self-healing loop also emits live SSE events to the frontend showing which phase and attempt is in progress.


Parametric System

Every generated model is parametric. The system:

  1. Defines parameters as named variables at the top of the generated script (e.g., body_width = 75.0)
  2. Saves a _parametric.py script with section markers used for rebuilding:
    python
    # ═══════════════════════════════════════ GEOMETRY GENERATION
    ...
    # ═══════════════════════════════════════ EXPORT
  3. Rebuilds on slider change via POST /api/rebuild — loads the saved script, replaces parameter values, re-executes the GEOMETRY section only, re-exports STL/STEP

This means parameter changes respond in 1–3 seconds with no AI usage.

Parameter Schema

json
{
  "name": "wall_thickness",
  "default": 2.5,
  "min": 0.5,
  "max": 10.0,
  "unit": "mm"
}

Product Library

product_library.py contains 98+ product templates with real-world reference dimensions. When a user prompt matches a product, its reference data is injected directly into the Claude context, ensuring accurate measurements.

Example Entry

python
{
  "keywords": ["iphone 15 pro", "iphone15", "iphone 15"],
  "name": "iPhone 15 Pro",
  "category": "phone_case",
  "dimensions": {"length": 159.9, "width": 76.7, "height": 8.25},
  "features": ["Dynamic Island", "USB-C port", "Action Button", "triple camera"],
  "notes": "Titanium frame. Camera island 37×30mm at top-left."
}

product_visual_knowledge.py adds per-category visual build guides with:

  • visual_profile — what the product looks like and key proportions
  • build_strategy — mandatory CadQuery approach (e.g., "use .revolve() + .spline() for mugs")
  • recognition_features — features that distinguish this category
  • position_map — where specific features are located

Optional Services

All optional services degrade gracefully — the app works fully without them.

MySQL Database

When DB_PASSWORD is set, the app stores:

  • Projects — named workspaces
  • Builds — each AI generation with code, parameters, file paths
  • Chat messages — full conversation history per project

Without MySQL, the app runs in-memory only (data is lost on server restart).

Setup:

sql
CREATE DATABASE product_builder CHARACTER SET utf8mb4;
CREATE USER 'cad_user'@'localhost' IDENTIFIED BY 'yourpassword';
GRANT ALL ON product_builder.* TO 'cad_user'@'localhost';

S3 Model Sharing

With AWS credentials set, use POST /api/s3/upload to generate a shareable URL for any model, and GET /api/s3/download/{key} to restore a shared model.

GLB Export (Three.js optimized)

With trimesh installed, POST /api/convert/glb converts STL to GLB format — smaller file size and faster loading in the 3D viewer.

Celery Async Tasks

With Celery + Redis running, use POST /api/build/async to queue builds and poll GET /api/task/{taskId} for results. Useful for server deployments with multiple concurrent users.


Frontend Architecture

Panels & Layout

The UI is a 3-column CSS grid: chat | resize-handle | preview. The columns are computed by getGridColumns() based on state:

  • Normal: ${chatWidth}% 8px ${100-chatWidth}% — user can drag the resize handle
  • Chat collapsed: 48px 0px 1fr — chat shows a vertical expand button
  • Preview collapsed: 1fr 0px 48px — preview shows a vertical expand button
  • Mobile (< 968px): 1fr with tab bar switching between panels

Component Tree

text
App.jsx  (layout shell)
├── header
│   ├── undo/redo buttons
│   └── projects button + quick export links
├── BuildProgressBar  (inline)
├── MobileTabBar  (inline, < 968px only)
├── main  (CSS grid)
│   ├── ChatPanel
│   │   ├── WelcomeScreen / message list
│   │   │   ├── UserMessage
│   │   │   └── AssistantMessage
│   │   │       ├── BuildingStatus (live steps + healing log)
│   │   │       ├── DesignSummary (explanation sections + downloads)
│   │   │       └── ErrorMessage (retry button)
│   │   ├── PromptInput  (upload badge when file active)
│   │   └── FileUpload  (drag-and-drop zone)
│   ├── ResizeHandle
│   └── PreviewPanel
│       ├── PreviewHeader (collapse btn + params toggle)
│       ├── MultiProductCanvas (Three.js)
│       ├── ParameterPanel (sliders)
│       └── ExportPanel (download + S3 share)
└── ProjectBrowser  (modal overlay)

State Management

AppContext.jsx provides a single useReducer store wrapping the entire app.

State Shape

typescript
{
  // Build
  status: 'idle' | 'building' | 'success' | 'error',
  result: BuildResult | null,
  messages: Message[],
  currentDesign: { code, parameters, explanation } | null,
  currentBuildId: string | null,
  currentProjectId: string | null,

  // Scene
  currentScene: Scene | null,
  sceneProducts: Product[],

  // Upload
  uploadedFile: UploadedFile | null,

  // Layout
  chatWidth: number,           // percent (20–80)
  isDragging: boolean,
  chatCollapsed: boolean,
  previewCollapsed: boolean,
  activeTab: 'chat' | 'preview',

  // UI flags

Key Actions

ActionEffect
BUILD_STARTSets status=building, resets progress
BUILD_PROGRESSUpdates progress bar value
BUILD_COMPLETESaves result, pushes history snapshot
UPDATE_BUILD_STEPUpdates SSE step in current assistant message
TOGGLE_CHAT_COLLAPSEDCollapses/expands chat panel
TOGGLE_PREVIEW_COLLAPSEDCollapses/expands preview panel
SET_UPLOADED_FILEStores upload metadata for NLP edit mode
UNDO / REDORestores previous/next history snapshot
REPLACE_LAST_PRODUCTUpdates 3D viewer after rebuild
RESET_WORKSPACEClears everything for a new project

Testing

Test files are at the project root. They use sys.path.insert(0, 'Backend') to import services directly (no pytest required — just run with Python).

bash
# Test parametric CAD generation
python test_phase32.py

# Test code completeness analysis
python test_completeness.py

# Test drone type detection
python test_drone_types.py

# Test phone case analysis
python test_phone_case_analyzer.py

# Test shape quality metrics
python test_shape_quality.py

# Test cq_warehouse integration
python test_cq_warehouse.py

Roadmap

  • Multi-model assembly — place multiple generated parts relative to each other and export as assembly STEP
  • Real-time collaboration — share a session link; multiple users edit the same design
  • Version history — git-like branching for design iterations
  • STL mesh editing — use mesh sculpting to NLP-edit mesh files (not just solids)
  • Material & finish metadata — annotate models with material, colour, surface finish for ordering
  • Direct 3D printer slicing — integrate Slic3r/PrusaSlicer API to preview G-code and print time
  • BOM generation — extract fasteners, bearings, and standard parts from cq_warehouse usage
  • Dimensions & annotations — overlay dimension lines and callouts on the 3D model
  • Drawing export — generate 2D engineering drawings (DXF/PDF) from the STEP model

Contributing

  1. Fork the repository
  2. Create a feature branch: git checkout -b feature/my-feature
  3. Follow the copilot-instructions.md for code conventions
  4. When adding new imports for AI-generated code, apply the triple-update rule:
    • Add to allowed_imports in _validate_code_safety()
    • Add to namespace in _execute_cadquery_code()
    • Add to namespace in rebuild_with_parameters()
  5. Add tests in test_*.py at the project root
  6. Submit a pull request with a clear description of what changed

License

MIT © 2026 Aslanmatejka

Parts

6
QtyValueFootprintReferences
4—MountingHole:MountingHole_3.2mmMH1, MH2, MH3, MH4
1—LED_SMD:LED_0805_2012MetricD1
1—RF_Module:ESP32-WROOM-32U1

Similar boards

  • 2 layers18.5 × 11.1 mmSwitch, the printed circuit board in 3D

    Switch

    byrantech/laptop

    anyon_e, a highly integrated, high end, open source laptop. Attempt the impossible.

    6 parts4.3kMIT
  • 2 layers14.6 × 33 mmBrain board SD Card 1, the printed circuit board in 3D

    Brain board SD Card 1

    rusefi/rusefi

    rusefi - GPL internal combustion engine control unit

    6 parts1.2k
  • 2 layers9 × 9 mmDpoint, the printed circuit board in 3D

    Dpoint

    Jcparkyn/dpoint

    Open-source digital stylus using camera tracking and inertial measurements

    6 parts1.1kMIT
  • 2 layers51.5 × 16.6 mmMicroswitchXY Endstop, the printed circuit board in 3D

    MicroswitchXY Endstop

    VoronDesign/Voron-Hardware

    Cool Hardware designed by the Voron Design group

    6 parts1.1kGPL-3.0
  • 2 layers21.9 × 76.7 mmAmiga External, the printed circuit board in 3D

    Amiga External

    hoglet67/RGBtoHDMI

    Bare-metal Raspberry Pi project that provides pixel-perfect sampling of Retro Computer RGB/YUV video and conversion to HDMI

    6 parts1kGPL-3.0
  • 2 layers88.9 × 50.8 mmBusinesscard, the printed circuit board in 3D

    Businesscard

    hackclub/OnBoard

    💡 Join 1,000 teenagers and make your first PCB with a $100 grant!

    6 parts915
  • 2 layers20.3 × 15.2 mmL0 RGB led, the printed circuit board in 3D

    L0 RGB led

    Luos-io/luos_engine

    Open-source and real-time orchestrator for cyber-physical-systems, to easily design, test and deploy embedded applications and digital twins.

    6 parts546MIT
  • 2 layers9 × 3.6 mmOpticalEncoder, the printed circuit board in 3D

    OpticalEncoder

    adamb314/ServoProject

    Getting the most out of your hobby servo

    6 parts545

Questions

How big is the Test pcb 001 PCB?

It measures 60 × 40 mm and has 2 copper layers, 1.6 mm thick.

What parts does the Test pcb 001 use?

6 parts. The full list of its parts is on this page, with links to each part's symbol and footprint where PCBKit's library has them.

Where can I download the Test pcb 001 KiCad files?

From the Aslanmatejka/inventa.ai repository on GitHub, at the commit this page was made from: the board is exports/pcb/test_pcb_001.kicad_pcb. PCBKit opens the board in the browser, with nothing to install.

Can I use the Test pcb 001 design in my own project?

Its repository states no licence, so its author keeps all rights: ask Aslanmatejka before reusing the design.