mz-optimize-memory
Pass
Audited by Gen Agent Trust Hub on Sep 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes untrusted data from the database environment to perform its functions. * Ingestion points: The agent reads the results of SQL queries, EXPLAIN plans, and system introspection views (SKILL.md). * Boundary markers: The instructions do not define specific delimiters to separate untrusted database output from the system prompt. * Capability inventory: The agent uses the
querytool to execute SQL commands which can alter the state of the database (SKILL.md). * Sanitization: There is no explicit logic described to sanitize or escape data retrieved from the catalog before it is processed. - [COMMAND_EXECUTION]: The skill interacts with the database using tools to execute queries and analyze performance. This behavior is mediated by strict operational rules, including the requirement to obtain human approval for any DDL or DML changes and the use of dedicated, non-production experiment clusters.
Audit Metadata