dax
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting and processing untrusted data, specifically DAX queries and model metadata provided by the user, which presents an attack surface for indirect prompt injection.\n
- Ingestion points: User-supplied DAX queries and semantic model metadata (such as measure definitions and relationships) are retrieved for analysis in Phase 1 of the optimization workflow (detailed in
references/dax-performance-optimization.md).\n - Boundary markers: Operational constraints are provided, such as instructing the agent to only modify measure/UDF definitions in the
DEFINEblock, which helps limit the scope of generated output but does not prevent the agent from being influenced by the input data.\n - Capability inventory: The skill uses tools (
te query,pbir) to execute DAX queries, profile performance, and read model metadata. It can also write and suggest modified DAX code.\n - Sanitization: The skill does not describe specific sanitization or escaping protocols for handling natural language that might be embedded within the processed DAX code or metadata.
Audit Metadata