mcp-sheet-filler: Skill for Claude Code

Purpose

The sheet-filler server manages objects (rows) with fields (columns). It prevents overwriting existing values, making it safe for incremental data collection.

Workflow

Setup (if needed)

Authenticate to Google Sheets

If not authenticated, use the device code flow:

# 1. Check status
filler_google_auth({ action: "status" })
→ { status: "Not authenticated. Use start_auth to begin authentication." }

# 2. Start auth
filler_google_auth({ action: "start_auth" })
→ { verification_url: "https://www.google.com/device", user_code: "ABCD-EFGH", device_code: "xyz123", instructions: "Visit ... and enter code: ABCD-EFGH" }

# 3. Tell user to visit URL and enter code, wait for confirmation

# 4. Complete auth
filler_google_auth({ action: "complete_auth", device_code: "xyz123" })
→ { status: "Authenticated to Google Sheets" }

Initialize a New Sheet

If the spreadsheet doesn't have data and fields tabs yet, create them:

filler_init()
→ { success: true, fieldsTab: "fields", dataTab: "data", keyField: "name" }

This creates both tabs with their header rows. Errors if either tab already exists.

Switch to a Different Sheet

To work with a different Google Sheet:

filler_use_sheet_id({ sheet_id: "1ABC...xyz" })
# or with full URL:
filler_use_sheet_id({ sheet_id: "https://docs.google.com/spreadsheets/d/1ABC...xyz/edit" })

Main Workflow

1. Understand the Schema

First, get the field definitions to understand what data to collect: filler_list_fields().

Fields with auto: true are candidates for auto-filling. Each field has:

• name - field identifier
• type - validation type (string, number, date, url, email, json, enum:...)
• instructions - how to collect this value
• example - example value

2. Get an Object

Two options for retrieving objects:

Option A: Get the next object with missing fields

filler_get_next_missing_fields_objects({ limit: 1 })

Returns objects with missing auto fields (default limit 1). If found: false, all objects are complete.