A framework to design sound, reproducible and scalable mining repositories studies on GitHub.
- 🧪 Reproducibility-first: declarative configuration and deterministic execution to enable repeatable experiments.
- 📈 Scalable: designed for large-scale repository mining studies on GitHub.
- 🧱 Soundness-focused: encourages transparent, bias-aware, and methodologically explicit study design.
- ⚙️ Modular: independent, reusable modules that can be composed into custom data-processing pipelines.
This project is written is Rust and requires Rust version 1.85. Install Rust by following the instructions on the official website.
Build Scyros from source:
git clone git@github.com:fxpl/scyros.git
cd scyros
cargo build --releaseThe binary is produced at target/release/scyros. You can optionally move it to a directory in your PATH for easier access.
To discover available commands and modules:
scyros --helpEach module provides its own usage documentation. For example, to inspect the module used to sample random repositories from GitHub:
scyros ids --helpSome modules interact with the GitHub API and require personal access tokens (PATs). Tokens can be created by following GitHub’s documentation: https://docs.github.com/en/github/authenticating-to-github/creating-a-personal-access-token.
Tokens must be provided as a CSV file passed via a command-line argument. The file must contain a single column named token, with one token per line:
token
fa56454....
hj73647.... GitHub enforces API rate limits. Using multiple tokens from the same account does not increase these limits. Users are expected to comply with GitHub’s API terms and rate-limit policies:
Scyros is introduced and described in the following large-scale empirical study. If you use Scyros in academic work, please cite:.
@misc{gilot2025largescalestudyfloatingpointusage,
title={A Large-Scale Study of Floating-Point Usage in Statically Typed Languages},
author={Andrea Gilot and Tobias Wrigstad and Eva Darulova},
year={2025},
eprint={2509.04936},
archivePrefix={arXiv},
primaryClass={cs.PL},
url={https://arxiv.org/abs/2509.04936},
} Gilot, A., Wrigstad, T., & Darulova, E. (2025). A Large-Scale Study of Floating-Point Usage in Statically Typed Languages. arXiv. https://arxiv.org/abs/2509.04936
This project is licensed under the Apache License 2.0. See LICENSE for details.