clickhouse-operations

Pass

Audited by Gen Agent Trust Hub on Feb 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill includes multiple scripts and examples that use powerful system commands for database administration.
  • The backup and restoration scripts in SKILL.md and examples/scaling-case-studies.md utilize sudo, systemctl, chown, and rsync to manage database services and data files.
  • Network diagnostic tools such as ping, mtr, nc, and telnet are included in troubleshooting guides to verify connectivity.
  • These commands are consistent with the skill's stated purpose for production management but require high-privilege access.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it instructs the agent to read and analyze data from untrusted sources within the database.
  • Ingestion points: The monitoring and diagnostic queries in SKILL.md and references/monitoring-queries.md read query text, error messages, and process information from system.query_log, system.processes, and system.errors.
  • Boundary markers: The instructions lack delimiters or explicit warnings for the agent to ignore instructions that might be embedded within the database logs (e.g., a malicious user could submit a query containing agent instructions).
  • Capability inventory: The agent has access to the Bash tool and is provided with examples for file system modification (rm, rsync), service control (systemctl), and network operations.
  • Sanitization: There is no evidence of sanitization or filtering of the content retrieved from ClickHouse system tables before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 24, 2026, 05:23 PM