creating-an-endpoint
Pass
Audited by Gen Agent Trust Hub on Oct 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided HogQL queries and variable definitions to configure API endpoints, presenting a standard attack surface for indirect prompt injection. The skill proactively mitigates this risk by providing specific security guidance.\n
- Ingestion points: User-supplied query strings and variable configuration payloads processed within the
SKILL.mdinstructions.\n - Boundary markers: The instructions recommend using structured variable syntax (e.g.,
{variables.x}) to clearly delineate user-controlled parameters from the static query logic.\n - Capability inventory: The skill utilizes specialized platform tools such as
endpoint-create,endpoint-run, andexecute-sqlto create, test, and manage data access points.\n - Sanitization: The skill provides explicit instructions to avoid unsafe patterns, such as exposing a
where_clausevariable that could lead to arbitrary SQL injection. It also details how enabling materialization serves as a security feature by enforcing that all declared variables must be passed, preventing unauthorized access to unfiltered datasets.
Audit Metadata