Point your agent at whatever you already run and get a plain-English deliverability audit in two minutes. It reads your real numbers, tells you exactly why you land in spam, and never touches a thing.
No card. No migration. 100% read-only, and nothing you connect is ever stored.
Broken DMARC. Mailboxes quietly over-sending. Inboxes stuck on a shared IP. Accounts that never warmed up. That's what kills deliverability, and it's invisible from every dashboard you already look at.
"Replies dropped." "We're in spam." "Bounces spiked." The dashboard shows the outcome, never the cause.
It reads the real signals: send volume, auth records, warmup age, IP type, and grades them against what actually works.
Every finding comes with the concrete next step, and it tells you when the fix is free and takes five minutes.
Example: connect your Smartlead account and the Doctor runs this whole flow, entirely read-only. The same pattern works for any connected tool.
A domain, an inbox, or your sequencer. Read-only.
Every inbox on the account, with its domain and status.
Daily send counts, ramp patterns, and activity per mailbox.
Warmup age, SPF/DKIM/DMARC, shared vs dedicated IP, placement.
Anything over the safe thresholds, with the exact fix.
It audits the platform you send from and the infrastructure your inboxes actually run on. No migration, no new account, just connect a key or point it at your domain.
Plus any mailbox over IMAP and any domain over DNS, no API key needed. New platform? One JSON file adds it, and the adaptive mapper learns the response shape on first contact.
Point it at a domain, an inbox, or a platform you already send from. It picks up your domains and runs the whole audit itself, read-only. You never touch a command.
Every platform returns a different shape. The adaptive mapper reads the real response, figures out the fields, and caches it, so it works even on APIs it has never seen.
~15 sends per mailbox per day, bounce under 3%, warmup 2 to 4 weeks, SPF/DKIM/DMARC, dedicated vs shared IP, the thresholds practitioners actually use.
Severity-ranked findings, each with the exact fix, and an honest "you're fine" when you are.
It runs entirely inside your own agent. There is no server on our side to send your data to. Whatever it reads is used for the audit in the moment, then it is gone.
We navigate you to the right set of tools and products for any kind of issue, matched to the exact problems the audit identifies.
Each finding names the real root cause, auth, warmup, volume, IP, or list hygiene, not just the symptom.
For every issue, we point you to the tool or product that solves it best, whichever fits your stack.
Some fixes are free and take five minutes. Where a product genuinely helps, we say so, and only there.
No signup. You never run a command. Connect what you use and it does the rest.
Paste this one-time install into Claude Code, on desktop or CLI.
Then just say:
Download the skill file, upload it to a new chat at claude.ai, and let Claude run the audit from there.
Free, open-source, and read-only. Run it on a live account without a second thought.