Readable Intelligence Skills & Demo Papers

From AI-generated mathematics to human understanding.

Website · Skill · Download · Demos · Contributing · Licensing

---

This is the public contribution repository for Readable Intelligence. It holds our paper-rewriting skill and seven demo paper pairs.

Current skill

Readable Intelligence Skill (Current Version) reorganizes proof-based mathematics and theoretical computer science papers for research readers. Statements and complete arguments are kept intact.

📖 Read	SKILL.md and its reference files
📦 Download	restructure-paper-qualified-revised.zip, the same file offered on the website
⚙️ Install	Unzip to get restructure-paper/, then install it as a skill in your AI tool
📝 Input	A manuscript

unzip downloads/restructure-paper-qualified-revised.zip   # → restructure-paper/

Demo papers

Each pair below is the original manuscript next to its Readable Version. Website publication is a separate release step.

Family	Paper	Original	Readable Version
102	The Unique Games Theorem	PDF	PDF
103	Exact Derandomization of Logarithmic Space: L = RL = BPL	PDF	PDF
107	An Upper Bound of 9/4 for the Matrix Multiplication Exponent	PDF	PDF
167	A power saving for planar unit distances	PDF	PDF
260	The spacetime Penrose inequality and enclosing area	PDF	PDF

The full upstream collection of 722 manuscripts lives at openai/math.

Contributing

You can propose a new skill, improve an existing one, or correct a demo by opening a pull request; CONTRIBUTING.md walks through the process. Maintainers review and merge accepted changes, then publish them to the website in a separate release step.

Attribution and licensing

• Demo manuscripts are under Apache-2.0, with attribution retained. Readable PDFs are modified editions, and each demo README says so.
• The skill was supplied without a separate license. LICENSES.md explains what that means for reuse and covers the full scope.