mirror of
https://github.com/supabase/supabase.git
synced 2026-10-10 11:55:05 +03:00
Charis
600a0ffcec
## I have read the [CONTRIBUTING.md](https://github.com/supabase/supabase/blob/master/CONTRIBUTING.md) file. YES ## What kind of change does this PR introduce? Docs (new Claude Code skill). ## What is the current behavior? There is no shared, written-down reference for the SQL safety model in Studio. The rules around `SafeSqlFragment`/`UntrustedSqlFragment`, sanitization utilities, and how to promote snippet content live only in code and contributor knowledge, which makes it easy for AI-assisted changes to bypass the type-based guarantees. ## What is the new behavior? Adds a `safe-sql-execution` skill under `.claude/skills/` that documents the proven-authorship security model: the three classes of SQL fragments, provenance tracking with branded types, sanitization utilities (`ident`/`literal`/`keyword`), the `acceptUntrustedSql` rule (event handlers only), and the special case that snippet content (`unchecked_sql`) must never be considered safe. Includes good/bad examples for the common patterns. ## Additional context <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit * **Documentation** * Added a comprehensive guide on secure SQL execution in Supabase Studio: explains provenance-based SQL safety, distinct categories of SQL fragments, how unsafe snippets must be explicitly promoted before execution, available sanitization helpers for user input, strict execution constraints to prevent accidental runs, and numerous examples demonstrating safe vs. unsafe usage and safe preview/runner patterns. <!-- review_stack_entry_start --> [](https://app.coderabbit.ai/change-stack/supabase/supabase/pull/46171?utm_source=github_walkthrough&utm_medium=github&utm_campaign=change_stack) <!-- review_stack_entry_end --> <!-- end of auto-generated comment: release notes by coderabbit.ai -->
