email-finder
Installation
SKILL.md
email-finder
The output is one verified work email — not a list of guesses. Sending to a guess burns the address (bounces hurt the sender domain's reputation; agentmail SES isn't infinite). If you can't reach high confidence on the address, don't send — fall back to a different surface (LinkedIn DM, GitHub issue, Twitter @-mention) or skip the target.
The workflow is always: find candidates → verify → send. Skip the verify step and you're back to guessing.
What input do you have?
The cheapest method depends on what you start with. Pick the highest row that fits:
| Input you have | Cheapest first move |
|---|---|
| GitHub handle | git log --format="%ae" → unique on a clone of any repo they push to (see Method 1) |
| LinkedIn profile URL | Crustdata person/enrich?linkedin_profile_url=... (Method 6) — single API call returns work email if available |
| Personal site / portfolio URL | curl + grep -E '[A-Za-z0-9._-]+@[A-Za-z0-9.-]+' for emails on the page itself or in the HTML source (often hidden but not encoded) |
| Conference/event speaker | The event's published speakers/program page often lists the email or the company; combine with Method 4 |
| Just a name + company | Permutator + verifier (Method 4); if company has 50+ employees, likely Apollo or Hunter is faster |
| Just a name | Stop. Find them somewhere first (LinkedIn, GitHub, Twitter, their company's team page). A name alone is too sparse. |