cotera

Airtable Data Cleanup Agent

Find duplicates, empty fields, and stale records hiding in your Airtable bases. Get a cleanup report with specific fixes, then let the agent apply them.

Data qualityDuplicate detectionRecord cleanupField validation

The Challenge

Your Airtable base started clean but months of manual entry have left it full of duplicates, records with missing fields, and rows nobody has touched in six months. You know the data quality is bad but cleaning it up means scrolling through hundreds of records, comparing fields, and deciding what to merge or delete. So it sits there, getting worse.

What This Prompt Does

Scan Schema

Reads the base structure to understand which fields are required and what types they expect

Find Issues

Searches for records with empty required fields, duplicate names or emails, and stale entries

Report

Generates a cleanup report grouped by issue type with specific records and recommended actions

Fix

Updates or deletes records on confirmation, merging duplicates and filling gaps

The Prompt

The Prompt

Task

Use @Airtable/Get Base SchemaName it "Airtable/Get Base Schema" and call it with @Airtable/Get Base Schema to understand the table structure and field types. Then @Airtable/Search RecordsName it "Airtable/Search Records" and call it with @Airtable/Search Records to scan for data quality issues: duplicates, missing fields, stale records, and inconsistent formatting. Generate a cleanup report and apply fixes on confirmation.

Input

The user provides:

  1. Their Airtable base ID
  2. The table name to clean up
  3. Which checks to run (or "all" for a full audit)

Example: "Scan my contacts table for duplicate emails and records with missing phone numbers"

Context

Schema Discovery

  1. Use @Airtable/Get Base SchemaName it "Airtable/Get Base Schema" and call it with @Airtable/Get Base Schema to read the table structure — field names, types, and which fields are required
  2. Identify key fields likely to contain duplicates (name, email, company, phone)
  3. Note field types to check for formatting inconsistencies

Data Quality Checks

Run whichever checks the user requests (or all of them):

Duplicates:

  • Search for records with matching name, email, or company fields
  • Flag potential duplicates with their record IDs and the matching field values
  • Recommend which record to keep based on completeness

Missing Fields:

  • Search for records where required fields are empty
  • Group by which field is missing and count affected records

Stale Records:

  • Search for records not modified in 90+ days
  • Flag records with no recent activity as candidates for archival or deletion

Formatting Issues:

  • Identify fields with inconsistent formatting (e.g., mixed phone number formats, inconsistent capitalization)
  • Suggest a standard format and list records that need updating

Applying Fixes

  • Use @Airtable/Update RecordName it "Airtable/Update Record" and call it with @Airtable/Update Record to fix formatting, fill in fields, and merge duplicate data into the kept record
  • Use @Airtable/Delete RecordName it "Airtable/Delete Record" and call it with @Airtable/Delete Record to remove confirmed duplicates or stale records
  • Always show the user what will change before making modifications
  • Process fixes in batches and confirm after each batch

Safety Guidelines

  • Never delete records without explicit user confirmation
  • When merging duplicates, always keep the more complete record
  • Show a preview of changes before applying them
  • Report how many records were affected after each operation

Output

Cleanup Report:

Duplicates Found:

  • Number of duplicate groups with the matching field
  • Record IDs and key field values for each group

Missing Required Fields:

  • Count of records per missing field
  • Sample record IDs for review

Stale Records:

  • Records not updated in 90+ days with last modified date

Formatting Issues:

  • Fields with inconsistent formatting and sample values
  • Recommended standard format

Recommended Actions:

  • Prioritized list of fixes with estimated impact

Example Usage

Try asking:

  • →"Scan my contacts table for duplicate emails and records with missing phone numbers"
  • →"Find all records in the leads table that haven’t been updated in 90 days"
  • →"Clean up the phone number formatting in my contacts table — make them all match (xxx) xxx-xxxx"