Set Up a Development Environment for Desktop Assistant

Getting a local development environment running for Desktop Assistant is straightforward. The project is a pure-Python application, so all you need is a compatible Python interpreter, a virtual environment, and the platform-specific dependency list provided in the repository. This page covers every step from forking the repo to running Jarvis in text-input (debug) mode so you can iterate on new commands without needing a microphone attached.

Prerequisites

Before you begin, make sure the following are available on your machine:

Python 3.9 or later — the project’s Ubuntu dependency snapshot was built against Python 3.9.7. Run python --version to check.
git — needed to clone the repository and manage branches.
A microphone (optional for development) — Jarvis uses speech_recognition to capture voice input. Set debug = True in config.ini to bypass the microphone entirely and type commands instead.

On Ubuntu you also need the system libraries portaudio19-dev and espeak before installing Python packages. The commands are shown in the setup steps below.

Setting Up Your Environment

Fork and clone the repository

Fork the repository on GitHub, then clone your fork:

git clone https://github.com/YOUR-USERNAME/DesktopAssistant.git
cd DesktopAssistant

If you plan to contribute back, also add the upstream remote:

git remote add upstream https://github.com/Harsha200105/DesktopAssistant.git

Create and activate a virtual environment

Create a virtual environment inside the repository root so your system Python is not affected:

python -m venv venv

Then activate it for your platform:

Windows
Ubuntu / macOS

venv\Scripts\activate

source venv/bin/activate

Your terminal prompt will be prefixed with (venv) once the environment is active.

Install dependencies

The repository ships two requirements files — one per platform.

Windows
Ubuntu

pip install --upgrade pip
pip install -r "Requirements&COC/requirements.txt"

The Windows requirements file installs:

pygame==2.0.1
SpeechRecognition==3.8.1
wikipedia==1.4.0
PyAudio==0.2.11
pyttsx3==2.90
tk

First install the required system libraries:

sudo apt-get update
sudo apt install portaudio19-dev python3-pyaudio
sudo apt-get install espeak

Then install the Python packages:

pip install --upgrade pip setuptools==58.0.4
pip install -r "Requirements&COC/ubuntu_requirements.txt"

Copy config.ini into the src/ directory

Jarvis reads config.ini from the current working directory at startup (see the os.path.isfile('./config.ini') check at the bottom of src/Jarvis2_4windows.py). Copy the template and then edit it:

cp "Requirements&COC/config.ini" src/config.ini

Open src/config.ini and fill in your details:

[DEFAULT]
master = YourName
search_engine = Google
debug = False
musicpath =
voice = Male
rate = 150
volume = 100
energy_threshold = 300

[EMAIL]
server = smtp.gmail.com
port = 587
username =
password =

Key settings:

Key	Description
`master`	The name Jarvis uses when greeting you
`search_engine`	`Google`, `Bing`, `DuckDuckGo`, or `Youtube`
`debug`	Set to `True` to use keyboard input instead of the microphone
`musicpath`	Absolute path to a folder containing `music1.mp3` … `music4.mp3`
`voice`	`Male` or `Female` — maps to the first or second pyttsx3 voice
`rate`	Speech rate in words per minute (default `150`)
`volume`	Volume as a percentage from `0` to `100` (default `100`)
`energy_threshold`	Microphone sensitivity; increase if Jarvis misses commands

Run Jarvis in debug mode

With debug = True in config.ini, Jarvis replaces the microphone with a simple input() prompt. This lets you test command routing without any audio hardware.

cd src
python Jarvis2_4windows.py   # Windows
# or
python Jarvis2.py            # Ubuntu

The Tkinter window will open and you will see the Command |--> prompt in your terminal. Type a command (e.g., wikipedia Python programming language) and press Enter to see Jarvis respond.

Code Structure

The project is intentionally lean — most logic lives in four Python files under src/:

src/Jarvis2_4windows.py

Windows entry point. Contains run() (reads config.ini, builds the take_command closure, calls wish_me) and main() (holds the phrases dict for zero-argument commands and the execute_the_command_said_by_user inner function for argument-bearing commands). This is the file to edit when wiring in a new command.

src/Jarvis2.py

Ubuntu entry point. A self-contained monolithic script that inlines speak, wish_me, take_command, and execute_the_command_said_by_user without importing from commands.py or actions.py. Useful as a reference for the expected end-to-end behaviour of each command.

src/commands.py

Individual command handlers. Each public function handles one user intent. Current handlers: command_wikipedia, command_whatsup, command_open, command_search, command_mail, command_nothing, command_hello, command_bye, command_play_music, command_pause_music, command_stop_music, command_unpause_music. New commands belong here.

src/actions.py

Core utilities shared across commands. Exports: speak(text), wish_me(master), open_url(url), search(search_query, search_engine), search_engine_selector(config), change_rate(query, take_command), change_voice(query, take_command), change_volume(query, take_command), set_gui_speak(command). The pyttsx3 engine is initialised at module level.

src/gui.py

Tkinter GUI. Builds a 700 × 500 root window with a scrollable Listbox that displays assistant responses and a Speak button wired to execute_the_command_said_by_user. The speak(text) function here appends Assistant: <text> to the listbox. mainloop is the Tkinter root.mainloop function re-exported for use by the entry points.

Requirements&COC/

Config and dependency files. config.ini — settings template; requirements.txt — Windows pip dependencies; ubuntu_requirements.txt — Ubuntu pip dependencies (also documents required system packages as comments).

Adding a New Command

Adding a voice command to Jarvis takes two steps: implement the handler in commands.py, then register the trigger phrase in Jarvis2_4windows.py.

1. Implement the handler in `src/commands.py`

Follow the existing pattern. Zero-argument commands accept no parameters; commands that need user input accept query and/or take_command (and optionally debug):

# src/commands.py

def command_joke():
    """Tell a random pre-written joke."""
    jokes = [
        "Why do programmers prefer dark mode? Because light attracts bugs.",
        "I told my computer I needed a break. Now it won't stop sending me Kit-Kat ads.",
    ]
    speak(random.choice(jokes))

For commands that need to search or open URLs, import and call the helpers from actions.py:

from actions import open_url, search, speak

2. Wire the phrase in `src/Jarvis2_4windows.py`

Zero-argument commands — add an entry to the phrases dict inside execute_the_command_said_by_user:

phrases = {
    "what's up":    command_whatsup,
    "nothing":      command_nothing,
    "abort":        command_nothing,
    "stop":         command_nothing,
    "hello":        command_hello,
    "bye":          command_bye,
    "play music":   command_play_music,
    "unpause":      command_unpause_music,
    "pause music":  command_pause_music,
    "stop music":   command_stop_music,
    "tell me a joke": command_joke,   # ← new entry
}

Also add the import at the top of the file:

from commands import (  # isort: skip
    ...
    command_joke,
)

Commands that need the query string — add an elif branch in the same function, after the phrases loop:

elif "joke" in query:
    command_joke()

Run Jarvis with debug = True after adding your command. Type the trigger phrase at the Command |--> prompt to verify routing before testing with microphone input.

Linting

The CI pipeline (.github/workflows/lint_python.yml) runs flake8 with a maximum line length of 88 and maximum cyclomatic complexity of 19. Run the same check locally before pushing:

flake8 src/ --count --max-complexity=19 --max-line-length=88 --show-source --statistics

The pipeline also runs bandit for security issues, isort for import ordering, codespell for typos, and mypy for type checking. Install them all at once:

pip install bandit black codespell flake8 flake8-bugbear flake8-comprehensions \
            flake8-return isort mypy pep8-naming pyupgrade safety

pyttsx3 initialises the sapi5 speech engine at import time on Windows (engine = pyttsx3.init("sapi5")). If you are running the CI linting steps on a Linux machine (as the workflow does), import errors from actions.py are expected and are handled by the --ignore-missing-imports flag passed to mypy.

Get Started

Configuration

Voice Commands

Contributing

Set Up a Development Environment for Desktop Assistant

Prerequisites

Setting Up Your Environment

Code Structure

src/Jarvis2_4windows.py

src/Jarvis2.py

src/commands.py

src/actions.py

src/gui.py

Requirements&COC/

Adding a New Command

1. Implement the handler in `src/commands.py`

2. Wire the phrase in `src/Jarvis2_4windows.py`

Linting

Build docs developers (and LLMs) love

Get Started

Configuration

Voice Commands

Contributing

Documentation Index

​Prerequisites

​Setting Up Your Environment

​Code Structure

src/Jarvis2_4windows.py

src/Jarvis2.py

src/commands.py

src/actions.py

src/gui.py

Requirements&COC/

​Adding a New Command

​1. Implement the handler in src/commands.py

​2. Wire the phrase in src/Jarvis2_4windows.py

​Linting

Build docs developers (and LLMs) love

Prerequisites

Setting Up Your Environment

Code Structure

Adding a New Command

1. Implement the handler in `src/commands.py`

2. Wire the phrase in `src/Jarvis2_4windows.py`

Linting