Built a free Python tool that tracks AI research papers from 28 tech companies on arXiv (MIT licensed)
Hey folks,
If you try to track what specific research labs (Google DeepMind, OpenAI, Meta, Anthropic, Apple, etc.) publish on arXiv, standard API queries usually fail. The official `<arxiv:affiliation>` XML tag only has about a 1% fill rate, and preprint aggregators like OpenAlex do not reliably index company affiliations for new preprints.
I built a lightweight open-source tool called paper-radar to solve this institutional tracking gap.
It queries arXiv across any date range you specify, matches author metadata against Research Organization Registry (ROR) identifiers and verified email domains, and isolates preprints from 28 frontier tech labs.
Key functionality:
- Splits lead authorship from bylines: Separates papers where the company lab was the primary driver from papers where a company researcher is merely an incidental co-author.
- Zero dependencies: Written entirely in Python standard library. No pip packages to install, no background daemons, no database, and no API keys required.
- Dual interface: Works as a standalone terminal script or as an installable Agent Skill for Claude Code, Codex, and standard skill.md agents.
- Git-friendly output: Emits clean, greppable Markdown summaries suitable for automated research digests or team notes.
The repository is open source under the MIT license:
tigerless-labs/paper-radar
Contributions and PRs adding verified institutional domains or lab aliases are very welcome. Let me know if you run into any parsing edge cases!
1 point
Mentions
Built a free Python tool that tracks AI research papers from 28 tech companies on arXiv (MIT licensed) Hey folks, If you try to track what specific research labs (Google DeepMind, OpenAI, Meta, Anthropic, Apple, etc.) publish on arXiv, standard API queries usually fail. The official `<arxiv:affiliation>` XML tag only has about a 1% fill rate, and preprint aggregators like OpenAlex do not reliably index company affiliations for new preprints. I built a lightweight open-source tool called paper-radar to solve this institutional tracking gap. It queries arXiv across any date range you specify, matches author metadata against Research Organization Registry (ROR) identifiers and verified email domains, and isolates preprints from 28 frontier tech labs. Key functionality: - Splits lead authorship from bylines: Separates papers where the company lab was the primary driver from papers where a company researcher is merely an incidental co-author. - Zero dependencies: Written entirely in Python standard library. No pip packages to install, no background daemons, no database, and no API keys required. - Dual interface: Works as a standalone terminal script or as an installable Agent Skill for Claude Code, Codex, and standard skill.md agents. - Git-friendly output: Emits clean, greppable Markdown summaries suitable for automated research digests or team notes. The repository is open source under the MIT license: tigerless-labs/paper-radar Contributions and PRs adding verified institutional domains or lab aliases are very welcome. Let me know if you run into any parsing edge cases!