postgres-readonly-queries
Pass
Audited by Gen Agent Trust Hub on Aug 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes data from external PostgreSQL databases, which creates a vector for indirect prompt injection if the database content is influenced by an attacker.\n
- Ingestion points: Data returned from SQL queries executed via
scripts/query.pyand displayed in the agent's context.\n - Boundary markers: The instructions do not include specific delimiters or warnings to the agent to ignore instructions embedded in the query results.\n
- Capability inventory: The skill allows for arbitrary SELECT queries and schema exploration on configured databases.\n
- Sanitization: Output is truncated to 10,000 rows and column content is capped at 100 characters, but no semantic filtering or sanitization of the data is performed to prevent instruction injection.
Audit Metadata