What problem does it solve? Business questions often get answered with hand-written SQL that guesses at joins and metric formulas, producing inconsistent results. This Skill routes analytics through a published Exasol Semantic Views model so metrics, dimensions, and compatibility rules come from the governed semantic layer instead of ad-hoc physical-table SQL. ## Core Features & Use Cases - Semantic Discovery: Query role-scoped SEMANTIC_AGENT views to list models, objects, metrics, dimensions, descriptions, and blocking validation errors before compiling anything. - Structured Compilation: Translate natural language questions into JSON requests compiled via COMPILE_REQUEST_JSON, with pre-checked metric/dimension compatibility and structured error or clarification responses. - MCP and Semantic SQL Paths: Execute governed semantic SQL through the official Exasol MCP Server preprocessor when EXECUTE SCRIPT is unavailable, and explain or record feedback on compiled requests using durable handles. - Use Case: An agent asked for "top 10 regions by revenue last quarter" discovers the sales model, verifies metric/dimension compatibility, compiles a structured request, executes the generated SQL under the caller's privileges, and records user feedback on the answer. ## Quick Start Ask the agent to answer a business question such as "show total revenue by customer region for completed orders" using the published Exasol semantic model.
