clickhouse-data-handling
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from ClickHouse system and user tables to perform monitoring and export tasks, which could theoretically allow maliciously crafted data in the database to influence the agent's behavior.
- Ingestion points: The agent is instructed to verify results against system tables and export user data from application tables in references/implementation.md (Steps 4 and 6).
- Boundary markers: The instructions do not include specific delimiters to separate data from instructions when the agent reads database outputs.
- Capability inventory: The skill uses Write and Edit tools to generate migration files and code, and the TypeScript examples perform database commands.
- Sanitization: The provided TypeScript implementation utilizes parameterized queries (query_params) to mitigate SQL injection risks.
Audit Metadata