For BOSC planning and review. Private to BOSC org committee (and a few privileged reviewers).
An AI-assisted pre-review tool for BOSC abstract submissions. It checks each abstract against four open source requirements:
| Rubric | What it checks |
|---|---|
| R1 — Available | Is the source URL accessible and does it contain real content? |
| R2 — Open | Does the repo have a recognized open source license that matches what the author declared? |
| R3 — Fresh | Is this project new to BOSC, or if previously presented, does the abstract describe meaningful progress? |
| R4 — Active | Does the project have multiple contributors, a history longer than a few days, and recent activity? |
- Claude Code installed and authenticated
-
Clone this repo and
cdinto it:git clone git@github.com:OBF/bosc-pre-review.git cd bosc-pre-review -
(Optional, for full R3 checking) Download the abstract PDF archive from the BOSC pre-review Google Drive folder and extract it:
mv Abstracts-BOSC2022-2025.tar.gz data/ cd data && tar xzf Abstracts-BOSC2022-2025.tar.gz && cd ..See
data/README.mdfor the expected directory structure.
From the root of this repo, start Claude Code and invoke the skill:
claude
/bosc-pre-review
Then paste in the abstract metadata. The expected format is a table with columns:
# Title License Source
These columns correspond to the abstract number, title, declared license, and source URL from the EasyChair submission form. Extra columns (e.g., author name, EasyChair boilerplate) are ignored.
The skill writes a tab-delimited .tsv file to the current directory with scores and comments for each rubric, suitable for pasting into a review spreadsheet.