EmbedGen

EmbedGen is an open source AI-assisted developer platform that converts natural-language requirements into embedded C firmware starter code, with additional outputs like pin maps, register notes, and circuit wiring suggestions.

---

Project Overview

Firmware bring-up often starts with repetitive setup work: peripheral initialization, pin planning, and first-pass control loop scaffolding. EmbedGen accelerates this phase by providing structured outputs from simple prompts.

Given inputs such as:

• target microcontroller (e.g., LPC2148, Arduino Uno, ESP32, PIC16F877A)
• task description (e.g., "temperature sensor using ADC")

EmbedGen returns:

• generated embedded C template code
• explanation of generated behavior
• pin configuration suggestions
• optional circuit wiring guidance

---

Feature List

• ✅ Firmware generation API (POST /generate-code)
  • Selects MCU + peripheral-oriented template from backend library
  • Returns C code, explanation, and pin configuration
• ✅ Circuit suggestion API (POST /generate-circuit)
  • Returns required components, pin connections, and wiring description
• ✅ Register explanation API (POST /explain-registers)
  • Provides concise register-level guidance from natural-language queries
• ✅ CLI interface (Typer)
  • generate and explain commands for terminal workflows
• ✅ Frontend dashboard (React + Tailwind)
  • Code Generator tab with syntax-highlighted code viewer, copy, and download actions
• ✅ Template and knowledge catalogs
  • MCU-specific firmware templates and circuit guidance in backend knowledge modules
• ✅ Smoke tests for critical API routes

---

Supported MCU Families

• LPC2148
• Arduino Uno
• ESP32
• PIC16F877A

---

Architecture

flowchart TD
  U[User / Developer] --> F[React Dashboard]
  U --> C[Typer CLI]
  F --> API[FastAPI Backend]
  C --> API

  API --> GEN[Generator Service]
  API --> CIR[Circuit Suggestion Service]
  API --> REG[Register Explanation Service]

  GEN --> TPL[Template Library\nbackend/templates]
  CIR --> KB[Circuit Knowledge Base\nbackend/knowledge/circuits]
  REG --> TPL

  API --> RESP[Structured JSON Response]
  RESP --> F
  RESP --> C

Loading

Repository Layout

/backend      FastAPI app, APIs, services, templates, circuit knowledge
/frontend     React + Tailwind dashboard
/cli          Typer CLI client
/examples     Embedded C sample files
/docs         Architecture and project documentation
/tests        Smoke and API behavior tests

---

Installation

1) Clone

git clone https://github.com/Pm-das/EmbedGen.git
cd EmbedGen

2) Create virtual environment

python -m venv .venv
source .venv/bin/activate

3) Install Python dependencies

pip install -r requirements.txt

4) Frontend dependencies (if running UI locally, using frontend/package.json)

cd frontend
npm install
cd ..

---

Example Usage

Start backend

uvicorn backend.main:app --host 0.0.0.0 --port 8000 --reload

Run tests

pytest -q

Lint and format checks

ruff check .
black --check .

Generate firmware code

curl -X POST http://localhost:8000/generate-code \
  -H "Content-Type: application/json" \
  -d '{
    "microcontroller": "LPC2148",
    "task": "Read ADC value and send via UART every second"
  }'

Generate circuit suggestion

curl -X POST http://localhost:8000/generate-circuit \
  -H "Content-Type: application/json" \
  -d '{
    "microcontroller": "LPC2148",
    "task": "Temperature sensor using ADC"
  }'

Explain registers

curl -X POST http://localhost:8000/explain-registers \
  -H "Content-Type: application/json" \
  -d '{
    "microcontroller": "LPC2148",
    "code": "U0LCR = 0x83; U0DLL = 97;"
  }'

---

CLI Examples

python cli/embedgen_cli.py generate "UART code for LPC2148"
python cli/embedgen_cli.py generate "ADC temperature logger" --microcontroller "ESP32"
python cli/embedgen_cli.py explain "U0LCR = 0x83; U0DLL = 97;" --microcontroller "LPC2148"

Using a custom backend URL:

python cli/embedgen_cli.py generate "PWM for Arduino Uno" --api-url http://localhost:8000

You can also set:

export EMBEDGEN_API_URL=http://localhost:8000

---

Screenshots (Placeholders)

Replace these placeholders with real screenshots from your local environment.

Code Generator

Code Debugger

Register Explainer

---

Contribution Guide

Contributions are welcome and appreciated.

Development workflow

1. Fork the repository
2. Create a feature branch

   git checkout -b feat/your-feature-name

3. Make your changes with tests/docs updates
4. Run checks locally

   pytest -q

5. Commit with a clear message
6. Open a Pull Request explaining:
   • motivation
   • implementation details
   • validation results

Contribution expectations

• Keep changes scoped and well-documented
• Prefer explicit, typed, readable code
• Add or update tests for behavior changes
• Update README.md / docs/ when interfaces change

---

CI/CD

• GitHub Actions workflow at .github/workflows/ci.yml runs linting (ruff), formatting (black --check), and tests (pytest) on every push and pull request.

---

Roadmap

Near Term

• Expand circuit knowledge coverage beyond ADC temperature sensing
• Add richer peripheral intent detection from natural-language tasks
• Improve API validation and error payload consistency
• Add CLI tests and frontend integration tests

Mid Term

• Add board profiles (clock assumptions, pin presets, toolchain hints)
• Add project export bundles (code + wiring + notes)
• Introduce user-configurable template styles (bare-metal vs HAL)
• Add authentication for hosted/multi-user deployments

Long Term

• Integrate compile/simulate feedback loops for generated firmware
• Support custom plugin packs for additional MCU families
• Add versioned generation history and collaboration workflows
• Integrate model-backed synthesis with confidence scoring

---

License

This project is licensed under the terms in LICENSE.