database-optimizer

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from database monitoring views such as pg_stat_statements and pg_stat_activity. These views include the full text of queries executed against the database, which may include comments or strings controlled by untrusted users. This creates a surface for indirect prompt injection where malicious instructions embedded in SQL comments could influence the agent's behavior during analysis.\n
  • Ingestion points: Ingestion of query text from pg_stat_statements in SKILL.md and pg_stat_activity in references/monitoring-analysis.md.\n
  • Boundary markers: The skill does not define delimiters or specific instructions to ignore content within the ingested query strings.\n
  • Capability inventory: The skill provides templates for executing powerful SQL commands including ALTER SYSTEM, CREATE INDEX, and ANALYZE.\n
  • Sanitization: There is no requirement or method provided for sanitizing the query text before it is processed by the agent.\n- [PRIVILEGE_ESCALATION]: The skill includes instructions for high-privilege database operations, such as ALTER SYSTEM SET (PostgreSQL) and SET GLOBAL (MySQL), which can modify global database configurations. Although these operations are standard for database optimization, they represent a significant increase in the agent's effective permissions within the database environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 11:50 AM
Security Audit — agent-trust-hub — database-optimizer