---
pretty_name: Repository Skill Miner (72k public-safe)
viewer: true
tags:
- datasets
- text
- code
- software-engineering
size_categories:
- 10K<n<100K
---

# Generated with Repository Skill Miner:

https://github.com/peytontolbert/repository-skill-miner

## Splits

- `train`: `74472` skill rows
- `repos`: `121` repo summary rows
- `annotations`: `29645` annotation rows
- `signals`: `2` revision signal rows

## Hugging Face configs

This dataset is pushed as multiple Hub configs:

- `default`: the main skill table
- `repos`: repo-level summary rows
- `annotations`: structured LLM annotation rows
- `signals`: revision-level signal rows


## Safety Mode

- `public_safe`: `true`
- Raw `snippet` and `doc_text` fields are excluded.

## Primary Columns

### train

- `repo_id`
- `miner_revision_id`
- `skill_id`
- `kind`
- `module`
- `qualname`
- `signature`
- `file_path`
- `line_start`
- `line_end`
- `annotation_summary`

### repos

- `repo_id`
- `miner_repo_name`
- `skill_count`
- `annotation_count`
- `signal_count`
- `annotation_models`

## Notes

- Default export strips raw code/doc text to reduce licensing risk for public release.
- If you need the exact code/doc strings for private workflows, rerun with `--include-code-text`.