sql-research
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (HIGH): The skill is highly susceptible to indirect prompt injection because it instructs the agent to retrieve untrusted data from the web and use it to perform write/execute-level operations.
- Ingestion points: External content retrieved via
WebSearchfor various SQL dialects and functions (defined inSKILL.mdresearch workflow). - Boundary markers: Absent; the instructions do not provide delimiters or warnings for the agent to ignore potentially malicious instructions embedded in the documentation it finds.
- Capability inventory: The agent is instructed to create/modify files (
research/*.md) and, more critically, to 'Implement SQL translations' and 'Write SQL-generating code'. - Sanitization: Absent; there are no verification or sanitization steps to ensure the retrieved SQL syntax or examples are safe before implementation.
Recommendations
- AI detected serious security threats
Audit Metadata