sql-injection-error
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by requiring the agent to ingest and process untrusted external data, such as database error messages provided by the user and engagement state from an MCP server. \n
- Ingestion points: The agent is prompted to ingest 'exact error messages' in the Assess step and calls 'get_state_summary()' to retrieve external state information. \n
- Boundary markers: There are no instructions defining delimiters or boundary markers to separate this untrusted data from the agent's core instructions. \n
- Capability inventory: The skill enables high-impact capabilities including shell command execution via tools like sqlmap and file system writes to the 'engagement/evidence/' directory. \n
- Sanitization: No input validation, escaping, or sanitization mechanisms are specified for the external content before it is processed by the agent.
Audit Metadata