PRIME-Community-Environments: Training-Ready RL Environments + Evals

Installation

Quick Installation (Recommended)

curl -sSL https://raw.githubusercontent.com/PrimeIntellect-ai/community-environments/main/scripts/install.sh | bash

sudo apt update && sudo apt install git-lfs

git lfs install

git clone git@github.com:PrimeIntellect-ai/community-environments.git
cd community-environments

git lfs pull

curl -LsSf https://astral.sh/uv/install.sh | sh
source $HOME/.local/bin/env

uv sync

uv run pre-commit install

uv tool install prime
prime config set-api-key <api-key>

Creating environments

Create a new environment template

prime env init <your-env-name> 

This will create an environment in environments/<your-env-name>. Enter the project directory with

cd environments/<your-env-name>

Then, edit your environment by implementing the load_environment function. To test, install the environment as a local package (editable) and then run the vf-eval entrypoint.

uv pip install -e .

uv run vf-eval <your-env-name>

Once you are done, push the environment to the registry.

prime env push 

If you bump the version in an environment's pyproject.toml, our CI will automatically build and publish that environment to the hub under the primeintellect organization. No manual action is required beyond the version bump.