Installation

Requirements

• Python 3.10 or higher
• pip

That's it.

Install

Recommended: Use a Virtual Environment

# Create a virtual environment (one-time)
python -m venv testgen-env

# Activate it
# On Windows:
testgen-env\Scripts\activate
# On macOS/Linux:
source testgen-env/bin/activate

# Install the package
pip install testgen-ai

Or Install Globally

pip install testgen-ai

That's it! The testgen command will work immediately after installation.

How it works: - On Windows: The first time you run testgen, it automatically adds Python's Scripts folder to your system PATH - On macOS/Linux: The command works immediately - Zero configuration required — everything happens automatically - On Windows, you may need to restart your terminal after the first use for the PATH to take effect in future terminals

Verify Installation

testgen --version

You should see the version number printed.

First time on Windows? The first run will automatically configure your system. You may see a terminal notification about PATH configuration — that's normal and expected.

---

Setting Your API Key (one-time setup)

Note: TestGen AI is currently in open development. Each user needs to supply their own LLM API key. We're planning to provide a managed, shared API key in a future release so setup is fully zero-config — but until that ships, a personal key is required to use the AI features. Getting one is free on most providers (Gemini has a generous free tier).

TestGen AI needs an API key for whichever LLM provider you want to use. Set it once using the built-in config command and it will work globally from any directory, forever:

# Google Gemini (default provider)
testgen config set GEMINI_API_KEY AIza...

# OpenAI
testgen config set OPENAI_API_KEY sk-...
testgen config set LLM_PROVIDER openai

# Anthropic / Claude
testgen config set ANTHROPIC_API_KEY sk-ant-...
testgen config set LLM_PROVIDER anthropic

# Ollama (local, no key needed — just set the provider)
testgen config set LLM_PROVIDER ollama
testgen config set LLM_MODEL llama3

Check what's currently configured:

testgen config show

Keys are saved to ~/.testgen/.env in your home directory and are picked up automatically by every testgen command from any project, on any directory.

Supported Providers

Provider	Key to set	Default model
Google Gemini	GEMINI_API_KEY	gemini-2.0-flash
OpenAI	OPENAI_API_KEY	gpt-4o
Anthropic	ANTHROPIC_API_KEY	claude-3-5-sonnet
Ollama (local)	(none)	llama3

Per-project override

If you need a different key or provider for one specific project, drop a .env file in that project's root:

OPENAI_API_KEY=sk-project-specific-key
LLM_PROVIDER=openai

Local .env always wins over the global config.

---

Non-Python Languages

For Python projects there is zero extra setup — pytest is already bundled.

If you're generating tests for JavaScript/TypeScript, Go, Rust, Java, etc., you need that language's runtime installed on your machine (Node.js, Go toolchain, cargo, JDK...). pip cannot install system-level language runtimes — this is the same requirement as just running any code in those languages normally.

UI / Browser testing (Playwright)

Playwright browser testing is an optional extra. If you need it:

pip install testgen-ai[browser]   # installs the Playwright Python package
playwright install                 # downloads browser binaries (one-time per machine)

The second step (playwright install) downloads Chromium/Firefox/WebKit binaries. This can't be automated by pip — it's a one-time manual step. For pure unit testing of Python code, you don't need this at all.

---

Troubleshooting

"testgen command not found" / "testgen is not recognized"

This rarely happens now since PATH is configured automatically, but here are solutions if it does:

Option 1: Run the command to trigger auto-setup

testgen --version

This runs the auto-setup if it hasn't run yet. Restart your terminal after.

Option 2: Use via Python module (always works, no setup needed)

python -m testgen generate ./src
python -m testgen config show
python -m testgen --version

This bypasses PATH entirely and works on any platform.

Option 3: Check if in PATH (Windows)

python -c "import sys; print(sys.prefix)"

This shows your Python installation directory. Add \Scripts to the end and manually add that folder to your Windows PATH if needed.

Option 4: Reinstall fresh

pip uninstall testgen-ai
pip install testgen-ai
testgen --version

Which option should I use?

• First time? Try running testgen --version — auto-setup should kick in
• Still not working? Use python -m testgen (always works immediately)
• Using a venv? Make sure it's activated with source testgen-env/bin/activate (or testgen-env\Scripts\activate on Windows) before running commands