spreadsheet-agent

Generates system prompts for agents that read and write spreadsheet data.

27.6k|2.7k|Updated Aug 6, 2024
One-click install
npx skills add https://github.com/mastra-ai/mastra --skill spreadsheet-agent
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: spreadsheet-agent
Source: https://github.com/mastra-ai/mastra/tree/main/packages/editor/src/ee/workspace/skills/spreadsheet-agent
Command: npx skills add https://github.com/mastra-ai/mastra --skill spreadsheet-agent

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Building agents that safely read and update spreadsheets (Google Sheets, Excel, CSV, Airtable, Notion databases) is error-prone: agents hallucinate successful writes, pick the wrong sheet, or run destructive operations without confirmation. This Skill provides an authoring playbook that produces a complete, safety-hardened system prompt for any spreadsheet automation agent.

Core Features & Use Cases

  • System prompt template: Generates a full agent prompt covering sheet selection, missing-input policy, decision rules, communication style, refusals, and completion criteria.
  • Safety guardrails: Enforces read-before-write, write verification via read-back, dry-run confirmation for destructive operations, and clean refusal when no spreadsheet integration is attached.
  • Tool selection guidance: Recommends attaching exactly one spreadsheet integration, plus optional date/time and scheduling tools for cadence-based workflows.
  • Use Case: A user asks for an agent that updates a Google Sheet of leads every morning; the Skill produces a 'Leads Sheet Updater' agent with a prompt that appends new rows, flags stale leads, verifies writes, and reports the exact sheet, tab, range, and row counts.

Quick Start

Ask the agent builder to create an agent that updates your Google Sheet of leads every morning, and this playbook will generate the complete system prompt with safety rules and verification steps.

Frequently Asked Questions about spreadsheet-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build an agent that updates Google Sheets automatically?

Define the target sheet, the update action, and the trigger (schedule or user request), then generate a system prompt using this playbook. The produced prompt enforces read-before-write, write verification, and a final receipt naming the sheet, tab, range, and row count.

How to prevent an AI agent from overwriting spreadsheet data?

Encode rules requiring the agent to read current values before writing and to never overwrite existing data without checking. For deletes, clears, or formula overwrites, require a dry-run with exact rows and ranges and stop for explicit user confirmation.

Which spreadsheet integrations does this agent playbook support?

The playbook covers Google Sheets, Microsoft Excel, CSV, Airtable, and Notion databases. It recommends attaching exactly one spreadsheet tool unless the goal is syncing data between two systems.

What happens if no spreadsheet tool is connected to the agent?

The generated prompt instructs the agent to refuse cleanly and explain that a spreadsheet integration is required. It names the missing connection rather than attempting or fabricating a write.

How does the agent verify a spreadsheet write succeeded?

The prompt requires a tool success response plus verification by reading back the affected range or using returned updated values. The agent must never claim a write succeeded until the spreadsheet tool confirms it.