DataUnmess.aiDataUnmess.ai
1

Create your key

Click the button below — it opens a new tab. Log in with your email and create your MCP key, then switch back to this tab and continue to step 2.

Login and create key
2

Choose your AI agent and follow its steps.

Pick your AI agent — Codex, Claude Code, Cursor, ChatGPT, Gemini, a custom app, or Manual — then follow the three steps. Using something that is not listed? Choose Manual for any other MCP-capable agent.

Codex Desktop

Copy this prompt and paste it in your AI agent.

Click the Copy button below, paste the prompt into your AI client, then send it.

Step 1 of 3

I already have a DataUnmess MCP key. Install DataUnmess MCP globally. Keep <YOUR_MCP_KEY> as the token placeholder.

Pastethen click send

Official Codex MCP documentation

Recommended next steps

Start with one of these recommended steps.

Once the setup is done, use one starter ask. We recommend starting with DataUnmess Memory onboarding so future MCP answers, research, dashboards, flowcharts, and data pipelines can reuse your company and KPI memory.

Seed DataUnmess Memory

Copy this after the setup shows Done.

Preview prompt
DataUnmess MCP start

Create a dashboard from a database or file

Copy this after the setup shows Done.

Preview prompt
Use DataUnmess.ai MCP to create a dashboard from my database or a file. Ask me which source to use, inspect the schema or columns, then build KPI cards, a trend chart, and the most useful breakdowns.

Filter a saved dashboard

Copy this after the setup shows Done.

Preview prompt
Use DataUnmess.ai MCP to add a transaction_date filter to my Sales dashboard for 2026-06-01 through 2026-06-30, and tell me which cards were updated or unsupported.

Import GA4 or HubSpot and schedule daily

Copy this after the setup shows Done.

Preview prompt
Use DataUnmess.ai MCP to import my GA4 data or HubSpot CRM data, create the standard dashboard, seed the last 7 days, schedule daily imports with a 7-day repair window, and ask before starting any longer backfill.

Clean up stale imports

Copy this after the setup shows Done.

Preview prompt
Use DataUnmess.ai MCP to list my imported datasets, identify any stale GA4 import tables, and delete only the one I confirm. Show the preview first and wait for my explicit yes before calling delete_dataset with the confirmation token.

Propose new insights and dashboards

Copy this after the setup shows Done.

Preview prompt
Use DataUnmess.ai MCP to review what you know about my company and available data, then propose new insights and dashboards that could be created. Group the ideas by business value and effort.

Report a DataUnmess bug

Copy this after the setup shows Done.

Preview prompt
A DataUnmess MCP tool appears to have failed because of a platform bug. Gather the exact tool call, expected vs actual behavior, error message, reproduction steps, environment, and a screenshot if possible, then call open_bug_report.
>With c:\tmp\sales.xls, create a DataUnmess bar chart of revenue by region
AI reads your file -> calls build_chart -> dashboard URL
>List my DataUnmess datasets and show the schema for sample-payments
Uses resources/list + resources/read
>In DataUnmess, show me the top 10 customers by payment_method
query_data with groupBy + orderBy + limit
>Build a DataUnmess dashboard with: KPI for total revenue, donut of revenue by payment method, line chart of daily orders
Multiple query_data + build_chart calls
>In DataUnmess, filter the Sales dashboard to transaction_date 2026-06-01 through 2026-06-30
set_dashboard_filter + dashboard re-query
>Create a DataUnmess flowchart of our order fulfillment: New Order -> Payment -> Approval check -> Ship or Reject
build_flowchart with labeled decision branches
>In DataUnmess, build a flowchart from this Mermaid: flowchart TD\nA[Start] --> B{Valid?}\nB -->|yes| C[Create]\nB -->|no| D[Reject]
build_flowchart accepts the `mermaid` field directly
>Use DataUnmess to analyze last month's sales and give me a written summary with highlights
query_data + build_analysis

Prepending "in DataUnmess" or "use DataUnmess MCP" signals to the AI that it should call the connected MCP server instead of answering inline.