The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes data from warehouse tables, views, and HogQL metrics to perform quality assertions, which represents a potential attack surface for indirect prompt injection.
Ingestion points: The agent reads data from system.information_schema tables (e.g., data_quality_checks, columns, data_quality_health) and the results of HogQL queries executed via posthog:execute-sql.
Boundary markers: The skill does not explicitly define delimiters for untrusted data, though it operates within the structured context of SQL/HogQL.
Capability inventory: The skill utilizes posthog:execute-sql, posthog:data-quality-check-create, posthog:data-quality-check-update, and posthog:data-quality-check-run to interact with the database.
Sanitization: The skill notes that saving a check validates the SQL composition, providing a layer of validation for authored assertions.
[DYNAMIC_EXECUTION]: The skill involves executing dynamic SQL and HogQL code via the posthog:execute-sql tool and the creation of custom_sql data quality checks. This is the primary intended functionality of the skill for data validation and warehouse management.
Evidence: Instructions guide the agent to take a compiled_query from check results and execute it directly using posthog:execute-sql to debug failures.

authoring-data-quality-checks