Quick Install

2 minutes to a working, offline installation.

Install

git clone https://github.com/dr-gareth-roberts/insideLLMs.git
cd insideLLMs
# Create and activate a virtual environment
python3 -m venv .venv
source .venv/bin/activate  # Windows: .venv\Scripts\activate

# Install the checkout (release publication is not yet established)
python3 -m pip install .

Using uv? Replace the last command with uv pip install .. PyYAML and Pydantic are included; validation works without development extras.

Verify

insidellms --version
# insidellms 0.2.0

# Offline smoke test: no API key or provider SDK required
insidellms quicktest "What is 2 + 2?" --model dummy

Create an Offline Harness

Run this from the directory where you want to keep the example. The command creates both harness.yaml and data/harness_dataset.jsonl; keep the config in that directory so its relative dataset path resolves correctly.

insidellms init harness.yaml --template harness
insidellms harness harness.yaml --dry-run

This generated DummyModel harness is the portable starting point whether you installed from a built wheel or from a source checkout.

Extras (Optional)

python3 -m pip install ".[openai]"       # OpenAI provider
python3 -m pip install ".[anthropic]"    # Anthropic provider
python3 -m pip install ".[nlp]"          # Text processing
python3 -m pip install ".[visualization]" # Report dependencies

For development from a clone, use python3 -m pip install -e ".[dev]" from the repository root. The providers extra installs OpenAI, Anthropic, and Hugging Face support; other providers may require their SDK separately.

Troubleshooting

“command not found: insidellms”

  • Ensure your virtual environment is activated
  • Try: python3 -m insideLLMs.cli --version

pip installation fails

  • Upgrade pip: python3 -m pip install --upgrade pip
  • Check Python version: python3 --version (must be 3.10+)

Next

First Run →