Claude Code Talks to Powr BI скачать в хорошем качестве

Claude Code Talks to Powr BI

DESCRIPTION “In this demo, we’re going to: 1. Connect Claude Desktop to the Power BI Modeling MCP server. 2. Use Claude to analyze a Store Sales Power BI report and output a full PDF style conversion plan for Databricks. 3. Then connect Claude for VS Code to the same MCP server and generate a structured set of instructions (model_instructions_plan.md and CSVs) that can drive downstream processes like Databricks dashboard creation.” _________________ Resources Below _________________ PBI Modeling MCP Server Config for Claude { "mcpServers": { "powerbi-modeling-mcp": { "type": "stdio", "command": "C:\\Users\\UserAcct\\.vscode\\extensions\\analysis-services.powerbi-modeling-mcp-0.1.9-win32-x64\\server\\powerbi-modeling-mcp.exe", "args": ["--start"] } } } Claude Desktop Prompt Power BI → Databricks Dashboard Conversion Analysis Instruction: Analyze the currently open "Store Sales" Power BI semantic model and report completely for Databricks dashboard conversion. Extract and report ALL of the details below. ________________________________________ 1. DATA MODEL • List ALL tables with row counts and key columns • Primary keys and relationships (many-to-many included) • Data source types (SQL, Excel, etc.) and connection modes • Import vs DirectQuery tables ________________________________________ 2. MEASURES & DAX • ALL measures with their full DAX code • Measure groups / display folders • Hidden measures • Complex time intelligence patterns (YTD, YoY, etc.) ________________________________________ 3. REPORT VISUALS (Page-by-Page) For each report page, document: Page N: [Page Name] Detail Description Visual type + title For each visual on the page X / Y axis fields Column or measure bindings Filters / slicers used Visual-level and page-level Sort order Default sort column and direction Conditional formatting Rules, thresholds, color scales Repeat for all pages. ________________________________________ 4. FILTERS & SLICERS • Page-level filters • Report-level filters • Slicer visual configurations (type, single/multi-select, default values) • Sync slicers across pages (Yes / No) ________________________________________ 5. DATABRICKS MAPPING PLAN For each major visual and measure, provide a mapping: Power BI Element Databricks Equivalent PBI Visual / Measure Databricks SQL widget type DAX formula Optimized SQL translation Table joins Unity Catalog 3-part name (catalog.schema.table) Filters Dashboard parameters ________________________________________ 6. DEPLOYMENT-READY ARTIFACTS Generate all of the following: 1. Lakeview Dashboard JSON — CREATE DASHBOARD spec 2. Parameter definitions — for all slicers 3. SQL queries — for the top 3 most complex visuals 4. Unity Catalog table references — full DDL with 3-part names 5. AI/BI Genie knowledge base — natural-language instructions for the dataset ________________________________________ Formatting & Priority • Output as Markdown with clear section headers • Prioritize: sales KPIs, store performance, and time-based analysis • Assume typical store-sales analytics patterns where report-level details are not directly accessible from the model API Claude VSCODE Prompt You are connected via VS Code to the Power BI Modeling MCP server and have read access to the currently open Power BI desktop semantic model. In plan mode, generate a step‑by‑step plan to: Inspect the semantic model and extract: All tables, columns, relationships, measures, calculated columns/tables, DAX expressions, roles, hierarchies, and display folders. Any embedded SQL or DAX queries and incremental refresh/partition logic that is exposed. Organize the extracted information into a single Markdown plan (model_instructions_plan.md) that will later drive the Databricks MCP server. Save this Markdown plan file in the currently open PBI-CLAUDE workspace/folder in VS Code (do not create a new root path; just place the file inside the existing PBI-CLAUDE folder). Briefly outline how this plan will be used to generate concrete files (e.g., CSVs for tables, columns, measures, relationships, roles, and a final Markdown instructions file) under the same PBI-CLAUDE folder when you switch out of plan mode. Specify how the Power BI semantic model’s logic (grain, relationships, measures, security) should be translated into Databricks Delta tables, Unity Catalog objects, and governed measures/views. Output only the plan (no code, no file content), as a clear numbered / bullet list of high‑level steps Claude will follow when switching out of plan mode.