supabase-common-errors
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external error data from Supabase, creating a surface for indirect prompt injection.\n
- Ingestion points: The agent reads
error.message,error.details, anderror.hintfrom Supabase SDK responses inreferences/implementation.md.\n - Boundary markers: Absent. No delimiters or instructions to ignore embedded commands are present in the provided diagnostic code.\n
- Capability inventory: The skill has access to
Read,Grep, andBash(limited tocurl,supabase,npx) tools as defined inSKILL.md.\n - Sanitization: Absent. No sanitization or escaping of the ingested error properties is performed in the diagnostic walkthrough code.\n- [DATA_EXFILTRATION]: The skill documentation suggests diagnostic commands that facilitate access to sensitive environment variables.\n
- Evidence:
references/examples.mdcontains the commandenv | grep SUPABASE. This command accesses sensitive credentials such asSUPABASE_SERVICE_ROLE_KEY, which provides administrative access to the database and bypasses security policies.
Audit Metadata