altinity-expert-clickhouse-caches
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION] (SAFE): The skill performs read-only operations on ClickHouse system tables (
system.events,system.parts,system.asynchronous_metrics,system.query_log). No DDL/DML or shell command execution is present. - [DATA_EXFILTRATION] (SAFE): The skill only accesses metadata and performance metrics (hit ratios, memory usage, table names). It does not access row-level application data or send information to external servers.
- [REMOTE_CODE_EXECUTION] (SAFE): No remote downloads or dynamic script execution identified. All logic is contained within static SQL queries.
- [INDIRECT_PROMPT_INJECTION] (LOW): While the skill processes table names and query logs, it does not interpolate this data into agent prompts in a way that could influence control flow. Use of boundary markers is recommended if summarizing query log output.
Audit Metadata