What this setup gives you
Datacooper uses two engines together. cwtwb generates .twb/.twbx dashboards, and cwprep generates .tfl/.tflx Prep flows.
The key business value is simple: your team gets real files that open directly in Tableau products, instead of temporary screenshots.
Step 1: Install and connect MCP tools
Install both tools in your AI client so prompts can call the generators directly.
# cwtwb
uvx cwtwb
# cwprep
uvx --from "cwprep[mcp]" cwprep-mcpStep 2: Build dashboard files with cwtwb
Use guided authoring for stable delivery: confirm datasource, choose chart primitives, assemble dashboard zones, then export.
- Core charts: Bar, Line, Area, Pie, Map, Text
- Advanced charts/actions: Dual Axis, Heatmap, Scatterplot, URL/Filter actions
- Recipe charts: Bullet, Butterfly, Lollipop, Calendar, Donut, Bump
Step 3: Build Prep flows with cwprep
Model your cleaning pipeline as nodes, then export to .tfl or .tflx. Keep logic transparent with SQL translation when review is required.
- Inputs: SQL/Table/Excel/CSV/CSV union
- Transforms: join, union, filter, calculate, pivot/unpivot, aggregate, rename, type change
- Output: publish-ready datasource config for Tableau workflows
Step 4: Review before rollout
For marketing demos and pilot projects, this checklist is enough.
- Open generated
.twb/.twbxin Tableau Desktop and confirm layout - Open generated
.tfl/.tflxin Tableau Prep and run once end-to-end - Store prompt + output version together for traceability
Starter prompt you can reuse
Build a weekly sales dashboard pack and a prep flow.
Dashboard side (cwtwb):
- KPI cards: Sales, Profit, Margin
- Trend: weekly sales line
- Breakdown: category bar chart
- Add one filter action from category to trend
Prep side (cwprep):
- Load orders + customers
- Remove invalid rows
- Unpivot monthly columns
- Add margin calculation
- Output final datasource files