agentleFS
manypicommanypicom/sales-skills/skills/email-verification/SKILL.md

email-verification

Skill1 starsChanged 2 months agoMITView on GitHub

Verify email addresses before sending so bounces never damage a sending domain, and interpret each verification status correctly. Use whenever the user mentions email verification, email validation, bounce rate, hard bounces, invalid emails, catch-all domains, disposable addresses, MX records, list cleaning, or asks whether an address is real. Also use before launching any cold email campaign, and when a campaign has been paused for bounces.

  • Sends data out

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask any provider whether they run a real mailbox probe.

    line 31
  2. Distinctive:

    Before every campaign: Always. No exceptions.

    line 54
  3. Distinctive:

    Read by whoever is on rota, or by nobody

    line 90
  4. Very specific:

    If bounces have already spiked, stop sending immediately — continuing makes it worse rather than working through it.

    line 107

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Email Verification

Verification exists for one reason: mailbox providers treat a high bounce rate as evidence that you're sending to a list you didn't earn. Cross the threshold and your domain's reputation drops, which affects every email you send from it, including the ones to real customers.

The threshold is lower than most people assume. Keep hard bounces under 2%, and treat 3% as an emergency. Some providers start filtering well below that.

Verification is cheap. A damaged sending domain takes weeks to recover and sometimes never fully does.

How verification actually works

A real verification is a staged process, and each stage is cheaper than the next.

1. Syntax check          Is it a well-formed address?           Free
2. Domain and MX lookup  Does the domain exist, and does it     Free
                         have a mail server at all?
3. Disposable check      Is it a throwaway-address provider?    Free
4. Role account check    info@, sales@, support@                Free
5. SMTP probe            Connect to the mail server and ask     Costs money
                         whether the mailbox exists, without
                         delivering anything

The SMTP probe is the part that matters and the part cheap tools skip. Some services guess from pattern databases — "firstname.lastname@ is usually right at this company" — and report a confidence score. That's not verification, and the bounce rate reflects it.

Ask any provider whether they run a real mailbox probe. If the answer is a score rather than a status, it's a guess.

Reading the statuses

StatusMeaningSend?
validThe mailbox exists and accepts mailYes
invalidThe server rejected the addressNever
no_mxThe domain has no mail serverNever
disposableThrowaway-address providerNever
catch_allThe domain accepts everything, so the specific mailbox can't be confirmedUsually yes
unknownThe probe was inconclusive — greylisting, timeout, blockingJudgement

The two at the bottom are where people make expensive mistakes in both directions.

catch_all covers a large share of real B2B domains, particularly larger companies. Discarding all of them removes a substantial part of a good list to avoid a modest number of bounces. Send to them, but treat them as a separate risk pool. See catch-all-risky-emails.

unknown usually means the receiving server declined to answer, which is a property of their configuration rather than of the address. Retry once after a day. If it stays unknown, decide by volume: a handful is fine to send, a thousand should be split off and sent slowly on a separate day.

The strict-versus-permissive choice is a real trade with no universally correct answer. Being strict costs you real prospects. Being permissive costs you reputation. If a domain is new or already damaged, be strict. If it's established and healthy, be permissive.

When to verify

  • Before every campaign. Always. No exceptions.
  • On any list older than 90 days. Roughly 2 to 3% of B2B addresses go bad each month through job changes.
  • On import, for anything from a form, a purchased list, an event, or a partner.
  • Not repeatedly on the same address. Cache results and re-check on age, not on every send.

Re-verify before re-engaging an old list. A list from last year is a bounce incident waiting to happen and it's the single most common cause of a domain getting burned.

Cost control

Verification is priced per probe, so avoid paying for probes you don't need:

  • Deduplicate first. Lists routinely contain 10 to 20% duplicates.
  • Run the free stages first. Syntax, MX and disposable checks cost nothing and remove a meaningful share of any list before a single probe runs.
  • Cache and reuse results across campaigns and across the workspace.
  • Get an estimate before running a large batch, split into billable and already-known.
# Free pre-checks you can run yourself before paying for anything
# 1. Malformed addresses
grep -vE '^[^@[:space:]]+@[^@[:space:]]+\.[a-zA-Z]{2,}$' emails.txt

# 2. Domains with no mail server — these can never receive mail
cut -d@ -f2 emails.txt | sort -u | while read -r d; do
  [ -z "$(dig +short MX "$d")" ] && echo "no MX: $d"
done

# 3. Duplicates
sort -uf emails.txt | wc -l

Those three passes typically remove 5 to 15% of a raw list at zero cost.

Role accounts

info@, sales@, contact@, hello@, admin@. They verify as valid because they exist, and they behave differently from personal addresses:

  • Read by whoever is on rota, or by nobody
  • Much lower reply rates
  • More likely to be reported as spam, because the recipient has no relationship with you
  • Sometimes the only address available for small and local businesses

For enterprise outbound, exclude them. For local business outreach where they're often all that's published, include them but keep them in a separate campaign so their metrics don't contaminate your read on the rest.

Spam traps

A spam trap is an address that exists only to catch senders who didn't earn their list. Hitting one damages reputation disproportionately, and verification cannot detect them — a trap is a working mailbox and verifies as valid.

The only protection is not acquiring them, which means: don't buy lists, don't scrape indiscriminately, don't mail addresses that have been dormant for years, and remove anyone who hasn't engaged across several campaigns.

This is the strongest practical argument against purchased lists. They're the primary source of traps, and verification gives no warning at all.

After a bounce incident

If bounces have already spiked, stop sending immediately — continuing makes it worse rather than working through it. The recovery sequence is in bounce-recovery.

Don't do these

  • Don't send to a list you haven't verified. Everything else in outbound depends on the domain surviving.
  • Don't trust a confidence score in place of a real probe.
  • Don't discard every catch_all. You'd be removing a large share of legitimate B2B prospects.
  • Don't verify once and reuse for a year. Data decays monthly.
  • Don't send to invalid because "it's probably fine". It isn't, and the whole point of verifying is to act on the result.
  • Don't use verification as a substitute for list quality. A verified list of the wrong people gets ignored instead of bouncing, which is a different failure with the same outcome.

Doing it in the workflow

The practical problem is that verification usually lives in a separate tool from the list and the sending, so it becomes a manual step people skip under time pressure.

ManyPI runs a real SMTP mailbox probe from a dedicated sending host, writes the status back onto the lead, and uses it automatically — invalid, no_mx and disposable are blocked from sending, catch_all and unknown are allowed by default with a Skip unverifiable setting for teams that prefer strict. Results are cached and reused across the workspace, and a run can be priced before it's queued. Over MCP, or over the API:

claude mcp add --transport http manypi https://mcp.manypi.com/mcp -s user
curl -X POST https://app.manypi.com/api/leads/validate \
  -H "Authorization: Bearer mpi_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{ "lead_ids": ["…"], "estimate_only": true }'

Whatever the tool, the rule is the same: nothing gets sent to an address nobody checked.

Related skills

email-deliverability, catch-all-risky-emails, bounce-recovery, lead-sourcing, suppression-list-management, crm-data-hygiene

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.