Do not commit private or sensitive research material to this repository.
Keep the following out of public files:
- API keys, access tokens, private keys, passwords, and service credentials.
- Local machine paths, private server paths, and unpublished repository setup notes.
- Human-subject data, participant identifiers, raw EEG or neural recordings, consent material, and IRB documents.
- Unpublished numeric results, reviewer strategy, paper drafts under embargo, rebuttal plans, private experiment logs, checkpoints, datasets, and generated outputs.
- Zotero exports or paper maps that include private notes, unpublished annotations, or non-public lab strategy.
Use ignored private overlays such as .private/ for local memory that should not be published. When a private lesson is broadly reusable, distill only the public-safe workflow, check order, or eval prompt into skills/*/references/.
Run the repository validator before publishing changes:
make validateThe validator scans public text files for common secret patterns and local path leaks, checks required skill files, validates JSON evals, resolves local Markdown links, and blocks large generated files.
If you find a leaked secret, private dataset detail, human-subject material, or unpublished result in public files, remove it from the branch immediately and rotate any affected credential. For public reports, open a GitHub issue that describes the affected file path without copying the sensitive content.