database-testing
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION] (HIGH): The skill utilizes functions like
db.raw()for arbitrary SQL execution andmigrate()/rollback()for schema modification. These capabilities can be hijacked to perform destructive database operations if input parameters are not strictly validated.\n- [PROMPT_INJECTION] (HIGH): The skill is vulnerable to Indirect Prompt Injection (Category 8) due to its high-privilege capabilities combined with untrusted data ingestion.\n - Ingestion points: Parameters in
Task()calls (e.g.,schema,migration) are potential injection vectors from external sources.\n - Boundary markers: No delimiters or isolation instructions are present to prevent embedded commands in parameters.\n
- Capability inventory: Includes data deletion (
db.users.delete), schema rollback, and raw SQL execution.\n - Sanitization: No evidence of input validation, escaping, or allow-listing exists in the provided implementation.
Recommendations
- AI detected serious security threats
Audit Metadata