supabase-audit-functions
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external Supabase Edge Function endpoints, which constitutes an indirect prompt injection surface. This is an expected architectural feature for a security auditing tool.
- Ingestion points: External API responses from Supabase Edge Function endpoints (e.g.,
/functions/v1/[function-name]). - Boundary markers: The instructions do not specify explicit delimiters for encapsulating untrusted response data within the agent's context.
- Capability inventory: The skill utilizes network requests (curl) and local file-write operations across the audit workflow.
- Sanitization: Examples in the documentation demonstrate the redaction of sensitive data (e.g., email, data fields) before saving to evidence files.
- [COMMAND_EXECUTION]: The skill generates and logs
curlcommands as evidence for detected vulnerabilities. These commands are intended for documentation and manual verification by the user. - [DATA_EXFILTRATION]: Audit results and evidence, which may contain sensitive configuration data from the target Supabase project, are stored in local tracking files such as
.sb-pentest-context.jsonand.sb-pentest-audit.log. This behavior is consistent with the skill's primary purpose of security auditing.
Audit Metadata