reporails

30k Corpus

Name: 30k Corpus
Author: reporails

AI community

SOURCE https://github.com/reporails/30k-corpus

Description

28,721 GitHub repositories analyzed for AI instruction quality across Claude, Codex, Copilot, Cursor, and Gemini. Full dataset behind "The State of AI Instruction Quality."

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

30K Corpus: AI Instruction Quality Dataset

Dataset for ["The State of AI Instruction Quality"](https://medium.com/@cleverhoods/the-state-of-ai-instruction-quality-30k-repo-analysis-ce49c7667a57) — a corpus analysis of 28,721 GitHub repositories containing instruction files for five coding agents. Provided via [Reporails CLI](https://github.com/reporails/cli).

What's in this dataset

File	Records	Description
`repos.jsonl`	28,721	Per-project: agents configured, stars, language, license, topics
`stats_public.json`	1	Aggregate statistics backing every table in the article
`validation_key.csv`	2,814	Sample classifications: atom text, tool labels, source location

Data dictionary

repos.jsonl

Each line is a JSON object with:

Field	Type	Description
`full_name`	string	`owner/project`
`canonical_agents`	string[]	Agents detected after deduplication: `claude`, `codex`, `copilot`, `cursor`, `gemini`
`stars`	int	GitHub star count at collection time
`primary_language`	string?	GitHub-detected primary language
`license`	string?	SPDX license identifier
`archived`	bool	Whether repo is archived
`topics`	string[]	GitHub topics