Validate WhatsApp
/validate-whatsappCheck any phone number against WhatsApp through the SyncGTM MCP and get a true or false per number back, at 1 credit per check.
Overview
Validate WhatsApp is a SyncGTM MCP agent that takes one phone number in international format and returns whether that number is registered on WhatsApp. It validates numbers you already hold rather than finding new ones, so it turns a raw mobile column into two usable lists: reachable on WhatsApp, and not. Each check costs 1 credit, and the charge lands the same whether the answer is yes or no.
What it does
- 1
Takes one input: `mobile`, a single phone number in full international format with the country code first, such as +14155552671.
- 2
Returns a two-field result per number — the `mobile` you sent and `registered` as true or false. That is the whole payload.
- 3
Runs one number per call. There is no batch or limit parameter, so a 40-number list is 40 calls and 40 credits.
- 4
Does not return a name, carrier, line status, WhatsApp profile or message history. A false tells you the number is not registered, not that it is a landline or a bad number.
- 5
Does not find numbers. Use find_mobile_number (12 credits, takes a LinkedIn URL) when the mobile is missing — scrape_phones_from_website (0.5 credits) only returns office and switchboard lines published on a company site.
- 6
Costs 1 credit per number checked. A number sent without its country code cannot match and still bills the credit, so normalize the list first.
How to use it
- 1
Download the skill file — it saves as validate-whatsapp.zip.
- 2
In Claude, open Settings → Capabilities → Skills and upload the .zip (or unzip the validate-whatsapp/ folder into .claude/skills/ for Claude Code).
- 3
Connect the SyncGTM MCP server using the setup guide for your client. It is a browser sign-in, no API key to paste.
- 4
Paste your numbers or point the agent at the mobile column in a CSV, then type /validate-whatsapp. The agent normalizes every number to international format and quarantines rows with no country code before it spends a credit.
- 5
Set the batch cap — the one setting worth tuning. Say "stop after 40 and report credits used" so a long list stays predictable.
Use cases
WhatsApp-first outbound
Split a prospect list into numbers that are on WhatsApp and numbers that are not before the sequence goes live. The WhatsApp step then only fires on contacts who can actually receive it.
Channel choice before dialing
Decide per contact whether the rep calls, texts or sends a WhatsApp message. Reps stop guessing which channel a mobile supports.
CRM mobile hygiene
Flag every mobile in the CRM with a WhatsApp yes or no so ops can route the no rows to email or phone sequences instead.
Event and inbound follow-up
Badge scans and demo requests arrive with a mobile and no context on how to use it. Check the list the same day and message the registered numbers while the conversation is fresh.
GTM workflow examples
Reopen stalled pipeline on the right channel
- 1Run `check_job_change` on 25 stalled opportunity contacts, passing `profile_url` (2 credits each, 50 credits), and cut anyone who has left the account.
- 2Run `find_mobile_number` on the 18 still in seat, one call per `linkedin_url` (12 credits each), and record which lookups came back empty.
- 3Run `validate_whatsapp` on every number that came back (1 credit each) to mark each contact WhatsApp-reachable or phone-only.
- 4Run `check_credits` (free) and log the actual spend against the play.
Outcome: One revival sheet for the 25 contacts: leavers cut, every mobile that came back tagged WhatsApp true or false, and the empty lookups routed to email instead.
Clean an inherited CRM mobile column
- 1Normalize the 300-row mobile column to international format and quarantine every row with no country code — zero credits spent so far.
- 2Run `check_credits` (free), then run `validate_whatsapp` on the first 100 normalized numbers (100 credits) to establish the WhatsApp rate before committing to the rest.
- 3Run `scrape_phones_from_website` (0.5 credits per site) on the accounts quarantined for having no usable mobile, and mark what it returns call-only — those are switchboard lines, so do not send them back through `validate_whatsapp`.
- 4Write the `registered` value back to the CRM as a WhatsApp_OK field on the 100 checked rows and route every false row into the email sequence.
Outcome: A WhatsApp_OK flag on the first 100 rows, a measured hit rate that prices the remaining rows at 1 credit each before you commit, and the no-mobile accounts carrying a call-only switchboard number.
Same-day event follow-up in two channels
- 1Run `enrich_person` on the 60 badge scans (1 credit each) to attach current title and company to each capture.
- 2Run `validate_whatsapp` on the 60 booth-captured mobiles after normalizing them (1 credit each) to find who can be messaged today.
- 3Run `verify_email` (0.3 credits each) on only the contacts that came back `registered: false`, so the non-WhatsApp group gets a deliverable address instead of a bounce.
- 4Split the export into a WhatsApp day-one list and an email day-one list, each row tagged with the title from step 1.
Outcome: Two ready-to-send follow-up lists from a single badge-scan export, with every contact assigned exactly one channel before the booth closes.
Prompts
Paste any of these into Claude once the skill is installed and the SyncGTM MCP is connected.
Is +14155552671 on WhatsApp?I have pasted 40 mobile numbers below. Normalize every one to international format first, then check each against WhatsApp. Stop after 40 numbers and report the credits used.Run check_credits, then take the mobile column from crm-export.csv, drop anything without a country code, and check the first 50 numbers against WhatsApp. Return two lists: on WhatsApp, and not.Find the mobile number for linkedin.com/in/janedoe, then tell me whether it is registered on WhatsApp.Check these 25 numbers against WhatsApp, cap the run at 25 credits, and give me a table with number, registered true or false, and the channel I should use.Best practices
Always send the country code. +14155552671, not 4155552671 — a local-format number cannot be matched and the attempt still costs a credit.
Normalize and deduplicate the whole list in one pass before the first call. The same number in two CRM rows bills 2 credits for one answer.
Only run mobiles. The parameter is `mobile`; switchboards and desk lines are almost never registered and each one burns a credit anyway.
Cap the batch and ask for the split. "Stop after 40 and report credits used", then "return two lists: on WhatsApp and not" — a per-row narrative is not usable.
Never re-check a false inside the same run. The second call bills another credit for an answer you already have, and false only means not registered — it does not tell you the number is wrong or that it is a landline.
Reach for the right sibling: verify_email (0.3 credits) when email reachability is the real question, and find_mobile_number (12 credits) when you do not have the number yet — this agent only validates.