---
id: arckit
version: "1.0.1"
license: Apache-2.0
license_treatment: permissive
maintenance: aging
---

# arckit — Tools for working with the Abstraction & Reasoning Corpus (ARC-AGI)

License: permissive · Maintenance: aging · Downloads: 339.7K/mo

## What it is and what it does

Arckit is a Python toolkit for working with the ARC-AGI family of reasoning datasets. It provides three main capabilities: loading the datasets (ARC-AGI, ARC-AGI-2, and Kaggle competition variants) without requiring separate downloads; visualizing tasks and grids as colored terminal output or high-quality vector graphics via drawsvg; and scoring model submissions in Kaggle format with configurable evaluation parameters. The package wraps numpy arrays as Task and TaskSet objects that are indexable by task ID or position, and offers both a Python API and command-line utilities (arcshow, arcsave) for inspection and export.

The package depends on numpy for array handling, rich for terminal formatting, and drawsvg for vector graphics output. It supports Python 3.8 through 3.13 and is classified as production-stable. Dataset versions can be pinned to avoid changes during research, and the underlying data follows the Apache-2.0 license from the original ARC-AGI repositories.

Use it for:

- Load and iterate over ARC-AGI-2 training and evaluation tasks for building reasoning models
- Visualize specific tasks to terminal or file (PDF/SVG/PNG) for papers, documentation, or manual inspection
- Score model predictions on Kaggle ARC competitions using the official evaluation metric
- Pin dataset versions in research environments to ensure reproducibility across time
- Extract task grids as numpy arrays for custom analysis or feature engineering

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Load, visualize, and evaluate solutions on the ARC-AGI and ARC-AGI-2 reasoning datasets through Python API and command-line tools, with built-in support for multiple dataset versions.

Yes, if you are working with ARC-AGI datasets for research or competition. The package eliminates the friction of manual dataset downloads and provides well-integrated visualization and evaluation tools. The aging maintenance status (357 days since release) is acceptable given the stable codebase and active repository, but monitor for breaking changes if you pin to a specific version long-term.

## Install

pip install arckit
uv add arckit
poetry add arckit

## Installing arckit

Before you install: Low install friction with a pure-Python wheel and three lightweight runtime dependencies. Maintenance status is aging (357 days since last release), though the repository remains active with recent commits and no archived status.

License in practice: Licensed under Apache-2.0 (permissive), allowing commercial and private use with minimal restrictions; you must retain license and copyright notices in distributions.

Quickstart:

pip install arckit

import arckit
train_set, eval_set = arckit.load_data()
task = train_set[0]
task.show()

Verify before relying:

- Whether the package handles all edge cases in grid visualization across different terminal environments
- Performance characteristics when working with large submission files in score_submission()

## Package facts

- License: Apache-2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: aging
- Downloads: 339.7K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

arc agi dataset loader, abstraction reasoning corpus tools, arc task visualization, arc submission scoring, reasoning dataset benchmark, reasoning-benchmark, dataset-tools

[View on SkillFed](https://skillfed.io/packages/arckit) · [View on PyPI](https://pypi.org/project/arckit/)