Imported from Findymail/findymail-cursor-plugin (skills/email-verification/SKILL.md). Install upstream with npx skills add Findymail/findymail-cursor-plugin --skill email-verification. Copyright stays with the author.

Email verification and reverse lookup

Two jobs, one skill:

1. Verify: tell whether addresses that came from elsewhere (a CRM export, a spreadsheet, a signup form) will deliver, without sending anything.
2. Identify: put a name, a LinkedIn profile and optionally a phone number on an address that arrived without context.

Tools and cost

One address per call. There is no batch endpoint.

Verify a list

1. Prepare the list, free

• Extract every address, lowercase it, and deduplicate.
• Drop anything that is not syntactically an email (no @, spaces, missing TLD). Mark it malformed without spending a credit.
• Flag role accounts (info@, sales@, support@, noreply@) in a note column. They can still be verified, but the user usually wants to know.
• Skip addresses that Findymail produced. Results from find_email and generate_lead_list are already verified.

2. Confirm the spend

Cost is 1 verifier credit per remaining address. For more than 25 addresses, state the count and wait for a go-ahead.

3. Verify

Call verify_email once per address. The response carries verified (true or false) and the mail provider (for example Google or Microsoft). Record:

• valid when verified is true
• invalid when verified is false
• error with the message when the call fails, and keep going

4. Deliver

• File input: write <original-name>-verified.csv next to the input with the original columns plus email_status (valid, invalid, malformed, error), provider, and note. Never overwrite the original.
• Inline input: a markdown table.
• Close with counts per status and the number of verifier credits used. Recommend removing invalid and malformed rows before any outreach sequence.

Write the file every 20 rows on long lists. On resume, skip rows that already have a final status and retry only error.

Identify who is behind an address

1. Call reverse_email_lookup with the email.
   • with_profile: false (default) returns the LinkedIn URL for 1 credit.
   • with_profile: true also returns the profile (name, title, company) for 2 credits. Use it when the user needs more than the URL, since it saves a separate lookup.
2. A no-match is free. Personal addresses (gmail.com and similar) match less often than work addresses. Say so rather than guessing.
3. Reverse lookup says nothing about deliverability. If the goal is to reply or reach out, also run verify_email.
4. Phone: only when asked. Pass the returned linkedin_url to find_phone (10 credits per number found) after confirming the cost.

Clean a CRM export end to end

When the user hands over an export and wants it "cleaned":

1. Prepare and verify every address as above.
2. For valid rows missing a name or company, offer to run reverse_email_lookup with with_profile: true (2 credits each) and confirm the spend.
3. Deliver one file with both sets of columns and a summary of what changed.

Errors

• 402 "Not enough credits": stop, deliver partial results, point the user to https://app.findymail.com. Note that verifier and finder credits are bought separately.
• 429: pause and retry the same address.
• 401: ask the user to reconnect the Findymail MCP server in Cursor settings.

Example prompts

• "Verify the emails in signups.csv before I add them to the sequence"
• "Is jane.doe@acme.com still valid?"
• "Who is behind m.rossi@brembo.com? I got a reply and don't know them"
• "Clean this HubSpot export: check every email and fill in the missing names"