The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by providing code examples that interpolate external data directly into SQL query strings.
Ingestion points: The etlPipeline function in SKILL.md ingests data from extractFromPostgres(), and the streamInserts function ingests from dataSource.
Boundary markers: None present; the code does not include instructions to the agent to treat interpolated data as untrusted or to ignore embedded instructions.
Capability inventory: The skill utilizes clickhouse.query and clickhouse.insert to execute database operations based on the processed data.
Sanitization: Absent; the bulkInsertTrades function uses standard template literals (e.g., '${trade.id}') for SQL value insertion instead of parameterized queries or prepared statements, which is a common vulnerability pattern.

clickhouse-io