The Agent Skills Directory

[COMMAND_EXECUTION]: The skill facilitates the execution of HogQL queries through the execute-sql tool. This functionality is central to its purpose of performing data analysis and does not pose a security risk in this context.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, such as saved insights via insight-get and event records from the database. It interpolates these values into SQL templates. While this presents an indirect prompt injection surface, the risk is mitigated by the specific analytical scope and the intended use of the tool within a data platform.
Ingestion points: Event data from the database and configuration from insight-get (SKILL.md).
Boundary markers: Not explicitly defined in the SQL templates.
Capability inventory: execute-sql, query-funnel, and query-funnel-actors (SKILL.md).
Sanitization: None explicitly defined in the provided markdown instructions.
[SAFE]: The skill does not perform any unauthorized network operations, access sensitive local system files, or use any form of obfuscation. All external resource patterns are consistent with the 'posthog' vendor identity.

analyzing-funnel-correlation