db-query
Fail
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill facilitates arbitrary command execution by interpolating user-provided SQL strings directly into a shell command (e.g.,
pnpm db:query:local -- "<SQL>"). An attacker can provide a string that closes the quotes and appends malicious shell commands, such as"; rm -rf /; ", which would be executed by the system shell. - Evidence: The execution pattern
pnpm db:query:<env> -- "<SQL>"found inSKILL.md. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted user data (SQL queries) and utilizes it in a high-privilege execution context without sanitization or boundary markers.
- Ingestion points: User-provided SQL queries extracted from requests as described in
SKILL.md. - Boundary markers: None identified; the input is directly placed within shell command delimiters.
- Capability inventory: Shell command execution via
pnpmacross local, staging, and production environments. - Sanitization: No input validation or escaping mechanisms are defined in the instructions.
Recommendations
- AI detected serious security threats
Audit Metadata