coding-agent

coding-agent is a small CLI wrapper around Google GenAI (Gemini) to run prompt-based agents that may call out to local helper functions.
This repo provides an installable package with a coding-agent command-line entry point and a Python module coding_agent for programmatic usage.

⚠️ This project expects a Gemini API key in your environment (see Configuration).

Features

• Installable package via pip (supports editable installs for development).
• Console script coding-agent for quick prompt runs.
• Includes a call_function helper module so the model can request function execution.
• Designed to be small and extensible.

Quick install

From PyPI (when published):

pip install coding-agent

From local source (editable, for development):

# inside a virtualenv
pip install -e .

From local source (non-editable wheel):

python -m build
pip install dist/coding_agent-0.1.0-py3-none-any.whl

Configuration

Create a .env file in the project root (or set environment variables). At minimum set:

GEMINI_API_KEY=your_real_gemini_api_key_here

The package uses python-dotenv to load .env at runtime.

Usage

CLI

After installing (and activating your virtualenv if needed):

# Run the CLI; wrap your prompt in quotes
coding-agent "Write a Python script for calculation"

If coding-agent is not available on your PATH, you can run it with the module mode:

python -m coding_agent "Write a Python script for calculation"

Programmatic

You can call the core runner from Python:

from coding_agent.cli import run

# run returns None and prints output; you can modify run() to return values for tests
run("Write a small script that computes factorial", verbose=True)

Development

Recommended project layout: use src/ layout to avoid accidental packaging of extra top-level folders.

coding-agent/
├─ pyproject.toml
├─ README.md
├─ LICENSE
├─ src/
│  └─ coding_agent/
│     ├─ __init__.py
│     ├─ cli.py
│     ├─ __main__.py
│     ├─ config.py
│     └─ functions/
│        └─ call_function.py
├─ streamlit_app.py
├─ main.py
└─ tests/

Install development dependencies and editable install:

# inside a virtualenv
pip install -r requirements-dev.txt
pip install -e .

Made with 💖 by Swikrit and open for contributions