Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.kontext21.com/llms.txt

Use this file to discover all available pages before exploring further.

Effective Prompting

Clear prompts help Claude produce precise Excel operations with fewer iterations. Use the following techniques to guide the model.

Provide Context Up Front

  • Reference the worksheet and range you want to work with. Example: “On Sales Q4, summarise B2:E200.”
  • Mention any attachments and what to do with them: “Use the attached contract PDF to validate totals.”
  • If the workbook has multiple tables, describe the relevant columns.

Be Explicit About Outputs

  • Specify the desired result: “Create a new sheet named Summary with columns Region, Total Revenue, Growth%.”
  • Indicate whether you expect a formula, value write, or format change.
  • For charts, state the chart type and data orientation.

Encourage Tool Planning

  • Ask Claude to outline its plan: “List the steps you’ll take and the tools you’ll call before executing.”
  • This yields a preview of planned operations so you can approve or adjust before tools run.

Use Follow-up Prompts

  • After a tool executes, request verification: “Confirm the totals in C2:C10 match the PDF.”
  • If results are off, explain what should change. Claude adapts quickly when given corrective feedback.

Sample Prompt

You’re working in the `Pipeline` sheet. Clean up the table in A1:F500 by:
1. Removing rows where Status = "Closed Lost".
2. Filling blank Probability cells with the average of the column.
3. Highlighting deals above $250k in green.
Show the steps you’ll take and ask for approval before writing.

Common Pitfalls to Avoid

  • Ambiguous references – Avoid “this column” or “these rows” without selecting them; Claude relies on precise addresses.
  • Oversized requests – Break large transformations into chunks to stay within token limits and maintain readability.
  • Skipping approvals – Do not auto-approve unfamiliar tools until you trust them in your environment.
Effective prompting is the fastest way to get high-quality results while minimising rework.