skills/datex/skills/db-query/Gen Agent Trust Hub

db-query

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use the dxs CLI tool (e.g., dxs configuration get, dxs configuration upsert) and jq for managing configuration files. These operations are part of the standard development lifecycle for Datex Studio and are considered vendor-authorized actions.
  • [PROMPT_INJECTION]: The skill processes untrusted user input through parameters such as $filter and $orderby to build database queries, which identifies an indirect prompt injection surface.
  • Ingestion points: The getListFlow.inParams array, specifically $filter and $orderby, provides external data from the UI to the query logic described in references/flow-db-datasources.md.
  • Boundary markers: No specific delimiters or boundary markers are documented to isolate user-supplied inputs within the generated predicates.
  • Capability inventory: The skill utilizes the $db global handle to execute CRUD operations (read, add, update, and remove) on backend storage components.
  • Sanitization: The skill mandates the use of a fluent operator DSL (e.g., .equals(), .in()) rather than native JavaScript operators and recommends platform-provided helper flows to translate and sanitize inputs during query construction.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 01:11 PM
Security Audit — agent-trust-hub — db-query