Privacy & data handling.
Pre-launch disclosure draft: This page describes the implementation, not a final published privacy policy. The operator must review it before launch.
Local library
Your saved skills, collections and comparison state use browser storage. They are not uploaded by the site during normal use. Export and import are initiated by you.
Skill nominations
When intake is enabled, repository URLs, slugs, names, descriptions and categories are stored for human review. Avoid including personal or confidential information. Receipts identify submissions; no account is created.
Reports and moderation
Issue reports store a public skill ID, issue category, explanatory text and review decisions. Do not submit private information in the report text. Reports do not automatically remove a listing. The service does not require accounts to report issues.
Abuse prevention
Submission, report and editor login endpoints use keyed hashes derived from request IP addresses to enforce rate limits. Hosting systems may separately retain access logs. The system uses Supabase for operational database storage when configured.
Third-party sources
Public catalog listings use skills.sh metadata. During moderation and scheduled evidence collection, GitHub is queried for repository and SKILL.md metadata. The site may store a bounded source preview, check timestamp, blob SHA and repository activity. These checks are not security audits. External source links and CLI installs follow their own third-party terms and practices.
Before launch
A responsible operator, valid contact point, retention/deletion schedule, legal jurisdiction and final privacy policy are still required. Do not treat this draft as legal advice or launch approval.