clickhouse
Fail
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Fetches and executes the official installation script for the clickhousectl CLI from the ClickHouse website.
- [COMMAND_EXECUTION]: Utilizes shell commands to install tools, manage local ClickHouse servers, and execute SQL queries against cloud or local instances.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data returned from external database queries which could contain malicious content designed to influence the agent behavior. 1. Ingestion points: Output from clickhousectl commands in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Shell command execution via clickhousectl in SKILL.md. 4. Sanitization: The skill includes a 'Safe query practices' section advising result limits, schema discovery, and human confirmation for expensive operations.
Recommendations
- HIGH: Downloads and executes remote code from: https://clickhouse.com/cli - DO NOT USE without thorough review
Audit Metadata