Working with XLSX Tools in Rakenne

XLSX Tools — Excel superpowers inside Rakenne

The XLSX Tools skill lets you work with Excel spreadsheets directly from your Rakenne project. You talk to the agent in natural language; it creates and edits .xlsx files for you.

With this skill, you can:

• Create new workbooks with multiple sheets, formulas, and professional formatting
• Read and summarize existing spreadsheets so you do not have to click through every tab
• Update existing files (add sheets, change formulas, restyle tables) and save new versions
• Recalculate formulas and check for errors before you send a file to a client
• Convert between formats like CSV, XLSX, PDF, ODS, and HTML

You never write code. You describe the structure, logic, and formatting you want; the agent does the spreadsheet work.

When to use XLSX Tools

Use XLSX Tools when you need to:

• Design a model or template: budgets, forecasts, risk registers, project trackers, KPI dashboards
• Review a client’s spreadsheet and extract the important parts into a report or workspace notes
• Clean up or refactor a legacy file: rename sheets, standardize formats, reorganize logic
• Check that formulas are safe: no #REF!, #VALUE!, divide‑by‑zero, or other error cells
• Move data between tools: client sends CSV, you need a polished XLSX; or you want a PDF for final delivery

If the task is “move data around in cells, apply logic, and export something tidy”, XLSX Tools are usually the right choice.

Getting ready: files, projects, and the skill

Before you start:

• Create or open a project in Rakenne where you want the spreadsheet work to live.
• Upload any existing files (XLSX, CSV, ODS, etc.) you want the agent to use into that project’s workspace.
• Install and enable the XLSX Tools skill from the Skills panel or library.

Once the skill is installed, just talk to the agent in the project chat. It will automatically decide how and when to use XLSX Tools.

Core workflows

1. Create a new Excel workbook from scratch

Goal: Ask the agent to design and build a complete workbook for you.

Typical flow:

1. Describe the purpose
   • What is this workbook for (for example, “3‑year SaaS forecast”, “vendor risk register”, “project plan”)?
   • Who will use it, and how?
2. Describe the structure
   • Which sheets you want (for example, Assumptions, Calculations, Dashboard)
   • What each sheet should contain (columns, sections, tables)
3. Describe the logic
   • Which numbers are inputs versus calculated
   • Any rules (for example, “ARR = MRR × 12”, “margin = (revenue – costs) / revenue”)
   • Any checks or flags (for example, “highlight rows where risk score > 15 in red”)
4. Describe formatting and usability
   • Currency and percentage formats
   • Freezing header rows, filters, and column widths
   • Colors for inputs versus formulas versus totals
5. Let the agent build and refine
   • The agent generates an .xlsx file and tells you where it saved it.
   • Download and open it in Excel or another spreadsheet app, check if it matches your expectations.
   • Ask for adjustments (“rename this sheet”, “add a Notes column”, “add a scenario selector”).

Example prompt

“Create a new Excel workbook for a 3‑year SaaS financial model.

• Sheets: Assumptions, P&L, Cash Flow, Summary.
• Put all hardcoded inputs in Assumptions (pricing, growth, churn, costs).
• P&L and Cash Flow should calculate monthly results over 36 months.
• Summary should show key KPIs (ARR, net margin, runway) with clear totals and chart‑ready tables.
  Freeze header rows on all data sheets and make input cells visually distinct from formulas.”

2. Build on top of an existing spreadsheet

Goal: Take an existing client or legacy file and improve or extend it.

Typical flow:

1. Upload the existing XLSX file into your project workspace.
2. Tell the agent what to keep and what to change:
   • New sheets or tables to add
   • Columns to insert or remove
   • Formulas to standardize or fix
   • Formatting and consistency issues to resolve
3. Ask the agent to save to a new file
   • Prefer names like client-budget-updated.xlsx so the original stays untouched.
4. Download, review, iterate
   • Open the updated file, spot‑check formulas and layout.
   • Ask for follow‑up changes as needed.

Example prompt

“I’ve uploaded client-budget.xlsx.

• Add a new sheet called Summary that aggregates monthly totals by category from the existing sheets.
• Make sure all currency fields use a consistent currency format.
• Add a final ‘Checks’ section on the summary that flags if total expenses exceed total revenue in any month.
  Save the result as client-budget-updated.xlsx.”

3. Extract data from a spreadsheet for review or reuse

Goal: Read what is inside without manually clicking through every tab.

Typical flow:

1. Point the agent to the file you have uploaded.
2. Say how you want the data back:
   • Human‑readable tables in the chat
   • Structured, analysis‑ready tables or lists
   • Focused on specific sheets or ranges
3. Use the extracted data:
   • Paste tables into a report, memo, or workspace notes
   • Ask follow‑up questions (“which vendors are high risk?”, “which months have negative cash flow?”)

Example prompts

“Summarize the contents of risk-register.xlsx. Show each sheet as a table and highlight any rows where the risk score is above 15.”

“From sales-pipeline.xlsx, extract only the Opportunities sheet as a table here in the chat. Then tell me the total pipeline value by stage.”

4. Recalculate formulas and check for errors

Goal: Make sure a formula‑heavy workbook does not hide broken links or divide‑by‑zero issues.

Typical flow:

1. Upload the workbook (or use one already in the workspace).
2. Ask the agent to recalculate and validate:
   • Recalculate all formulas
   • Scan for error values
   • Report where problems appear (sheet and cell references, plus a short explanation if possible)
3. Fix issues iteratively:
   • Ask the agent to repair broken references, handle divide‑by‑zero, or add defensive logic.
   • Re‑run the check until there are no errors.

Example prompt

“For saas-model-v3.xlsx, recalculate all formulas and report any cells that contain errors like #REF!, #VALUE!, or divide‑by‑zero. Then suggest safe fixes for each issue, and create a cleaned‑up version of the file with those fixes applied, saved as saas-model-v3-fixed.xlsx.”

5. Convert between CSV, XLSX, PDF, ODS, and HTML

Goal: Move spreadsheets between tools and formats without re‑doing work.

Typical conversions:

• CSV → XLSX: Clean import, typed columns, better formatting
• XLSX → CSV: For systems that only accept text files
• XLSX → PDF: For sending read‑only versions to clients
• XLSX ↔ ODS: Moving between Excel and open formats
• XLSX → HTML: To embed tables in documentation or intranets

Typical flow:

1. Upload the source file (CSV, XLSX, ODS, and so on).
2. Tell the agent what you want:
   • Target format
   • Which sheets or ranges to include, if relevant
   • Any expectations (page orientation for PDF, basic styling, and so on)
3. Download the converted file from the workspace.

Example prompts

“Convert transactions.csv into a well‑formatted Excel workbook with a single sheet. Use appropriate column types (dates, numbers) and freeze the header row. Save it as transactions.xlsx.”

“Convert board-report.xlsx into a landscape‑oriented PDF suitable for printing, with gridlines hidden. Save it as board-report.pdf.”

Practical limitations and things to watch for

To set expectations correctly:

• Very large files
  • Extremely big workbooks (hundreds of thousands of rows or dozens of complex sheets) can be slow or may hit limits.
  • If you suspect this, tell the agent to work on a subset (for example, “only the last 24 months” or “only the Summary sheet”).
• Macros and VBA
  • Macros, VBA, and complex add‑ins are not executed. XLSX Tools work with data, formulas, and formatting, not custom scripts.
  • If a workbook’s logic depends on macros, describe that logic in plain language so the agent can re‑implement it with normal formulas.
• Exotic Excel features
  • Some advanced features (certain data connections, external links, very complex conditional formatting) may not round‑trip perfectly.
  • For important models, keep the core logic as plain formulas and tables.
• Layout and printing
  • The agent can set print areas, orientation, and basic layout, but very fine‑tuned print layouts might need a quick manual tweak in Excel.
• Overwriting files
  • Always ask the agent to write to a new file name (for example, _updated, _final, _report) instead of replacing the original. This makes it easy to compare and roll back.

Prompt patterns you can reuse

You can adapt these patterns to your own context.

Create a model

“Create a new Excel workbook for a 24‑month consulting project plan.

• Sheets: Timeline, Budget, Resources, Summary.
• Include start and end dates, phases, milestones, daily rates, and utilization.
• The Summary sheet should show total fees, margin, and a chart‑ready table of monthly revenue.
  Use clear formatting and freeze header rows.”

Refactor a messy file

“Take client-forecast-old.xlsx and:

• Standardize all currency columns to the same format.
• Move all assumptions into a dedicated Inputs sheet, linked from the other sheets.
• Add checks that flag if total expenses exceed total revenue in any quarter.
  Save the cleaned‑up version as client-forecast-clean.xlsx.”

Extract and analyze

“From incident-log.xlsx, extract the last 12 months of entries and show them as a table here. Then group them by severity and month, and give me a short analysis of trends.”

Validate before sending to a client

“For soc2-controls-tracker.xlsx, recalculate all formulas, list any errors you find, and suggest fixes. Once the errors are resolved, prepare a final version named soc2-controls-tracker-final.xlsx.”

Convert and polish

“Convert export.csv into a nicely formatted Excel workbook with filters on the header row and reasonable column widths. Save it as export-formatted.xlsx.”

Make the most of XLSX Tools

To get the best results:

• Be explicit about structure: mention sheet names, column names, and relationships between them.
• Separate inputs and calculations: ask for an Inputs or Assumptions sheet so changes are safe and localized.
• Describe conventions once: for example, “treat all currency as USD with two decimals” or “hardcoded inputs in blue, formulas in black”.
• Iterate with the agent: treat XLSX Tools as a spreadsheet assistant. It is normal to build and refine a workbook over several short prompts.