Lint Nurture Sequence

A nurture sequence underperforms when its emails drift apart — one has a bold clickable link above the image, the next only has the image (so anyone who blocks images has nothing to click); one signs off warmly, another just stops. This skill runs a deterministic check over the whole set and tells the operator exactly what's off, per email and across the set.

It's the quality gate between design-nurture-sequence (draft) and wire-nurture-sequence (ship) — and it works on a live queue too, to catch drift in something already running.

Step 1 — Pick the source

Two ways in:

• A live auto queue — lint what's actually deployed:

  python ~/.claude/bos-run.py tool lint-sequence --queue <auto_queue_id> --json

  (Get the queue id from /nurture-health or python ~/.claude/bos-run.py tool dump-crm-bundle --resources auto_queues.)
• Local drafts — lint before shipping, e.g. the drafts from design-nurture-sequence:

  python ~/.claude/bos-run.py tool lint-sequence --drafts drafts.json --json

  Drafts file: [{"label": "Day 0", "subject": "...", "body": "<p>...</p>"}, ...].

Useful flags: --signoff "Warmest regards, Sam" to match the operator's locked closer; --allow-em-dash if the operator is fine with em dashes (default flags them, because the house style avoids them).

Step 2 — What each check means

Per email:

Across the set:

• MIXED cta_above_image is a FAIL on purpose — some emails following the pattern and others not is the single biggest "half-built" tell.
• Inconsistent sign-offs or P.S. presence are WARNs — align them unless the variation is deliberate.

Step 3 — Present and route the fixes

Lead with the verdict and the fails, then the warnings, then "the rest are clean". Translate each flag into the concrete fix.

Sequence lint — 7 emails: 2 fail, 3 warn

✗ Day 7  — Want AI to build your CRM for you?
     ✗ cta_above_image: image has no text link above it — add a bold "Connect Claude here:" link before the image
✗ Day 14 — Watch Claude wire up your pipeline
     ✗ cta_above_image: same issue
⚠ Day 21 — An automation that runs forever
     ⚠ signoff: 'Warmest regards' not found

Across the set:
  ✗ MIXED: Days 0/49 have a text CTA above the image, Days 7/14/21/28/38 don't — make it consistent.

Fix path: hand the failing emails to /design-nurture-sequence to rewrite, then
/wire-nurture-sequence to push the corrected copy. Re-run this lint to confirm green.

For live-queue fails, the fix is to rewrite the copy (design-nurture-sequence) and re-wire it (wire-nurture-sequence) — this skill never edits the queue itself. It only reports.

What to never do

• ❌ Don't edit emails or the queue from this skill — diagnose and route only.
• ❌ Don't soften a MIXED cta_above_image to a warning — inconsistency across the set is the failure that matters most.
• ❌ Don't treat --allow-em-dash as the default — only pass it if the operator has said em dashes are fine in their voice.

Output shape

Open with the one-line verdict (N emails, X fail, Y warn). List failing emails with the concrete fix, then warnings, then the set-wide consistency findings. Close with the fix path (rewrite → re-wire → re-lint) and offer to kick it off.