supabase-data-handling
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and export records from Supabase database tables and storage buckets, which can contain untrusted content supplied by users that may influence agent behavior.\n
- Ingestion points: The
scanTableForPIIfunction inreferences/rls-and-pii.mdandexportUserDatainreferences/deletion-and-export.mdretrieve data directly from external PostgreSQL tables and storage buckets into the agent's context.\n - Boundary markers: The instructions do not define explicit boundary markers or delimiters (such as XML tags or specific 'ignore' headers) to separate untrusted data from the agent's core instructions.\n
- Capability inventory: The skill allows for significant command execution via the
allowed-toolsmetadata inSKILL.md, specificallyBash(supabase:*),Bash(psql:*), andBash(npx supabase:*).\n - Sanitization: Although the skill provides snippets for PII detection and redaction (e.g., in
references/implementation.md), it lacks logic to sanitize or filter for prompt injection instructions within the retrieved database content.
Audit Metadata