agentleFS
BenPremBenPrem/business-brain/.claude/skills/lead-flow-verifier/SKILL.md

lead-flow-verifier

Skill1 starsChanged 3 months agoMITView on GitHub

Fires marked test submissions through every lead entry point on a deployed site (forms, tel links, quiz funnels, booking, chat) and verifies the full chain — form-platform submission, notification email, <CRM> record, analytics events — as a PASS/FAIL/COULDN'T-VERIFY matrix. Triggers - "verify lead flow", "test the forms", "lead capture check". NOT pre-delivery QA — use site-qa-checklist.

  • Sends data out

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run this as a launch gate before declaring any client site live "with lead capture," and re-run after any form, function, or DNS change.

    line 18
  2. Distinctive:

    Test submissions trigger REAL notifications.

    line 24
  3. Distinctive:

    If the target is your inbox, confirm the email actually arrived.

    line 83
  4. Distinctive:

    <CRM>/automation record: ONLY for chains that claim one exists.

    line 86
  5. Distinctive:

    Delete each test submission via the platform API, then re-list to confirm it's gone — verify the delete too.

    line 110
  6. Distinctive:

    If a notification landed in the client's inbox, have the founder flag it as a test.

    line 114

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “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

Lead Flow Verifier

Doctrine: the user-facing read surface is the only ship contract. A form that renders, a 200 on POST, or a dashboard saying "connected" proves nothing — a lead flow "works" only when a real test submission is visible at every downstream step a human relies on. COULDN'T-VERIFY is a first-class result — report it honestly, never round it up to PASS.

Corollary: an empty read right after an async write is INCONCLUSIVE, never "failed." Form-platform APIs and notification pipelines lag by minutes. Confirm at the destination (the notification inbox, the <CRM> record, the function log) — and symmetrically, a 2xx/"success" transport response proves nothing landed either.

Run this as a launch gate before declaring any client site live "with lead capture," and re-run after any form, function, or DNS change. site-launch-cutover delegates its post-launch form check here.

Gate before firing anything

Test submissions trigger REAL notifications. If any land in the client's inbox, phone, or <CRM> (not yours), tell the founder exactly which entry points will fire and where notifications go, and get a go-ahead first — a mystery lead confuses the client. Standard payload, always clearly marked:

  • Name: TEST — <YOUR BUSINESS> verification
  • Email: your own address — Phone: the founder's cell (confirm it, don't guess)
  • Message: TEST submission from <YOUR BUSINESS> verifying lead capture — please ignore, we will remove it.

Phase 1 — Inventory entry points (from the DEPLOYED site, not local source)

curl -s every page of the live URL (or crawl sitemap.xml) and grep for:

  • Forms: <form, data-netlify, action=, honeypot fields, hidden form-name inputs
  • Phone/email: href="tel: (dedupe numbers), mailto:
  • Quiz/questionnaire funnels: multi-step JS — grep inline + linked JS for fetch(, XMLHttpRequest, or serverless-function paths to find the real submit endpoint
  • Booking: Calendly/Cal.com/appointment iframes or links
  • Chat widgets: third-party script tags

Diff the inventory against the local source repo. An entry point in source but missing live = deploy drift; live but not in source = someone edited the wrong copy. List every finding as a matrix row BEFORE testing any.

Intentionally-unwired demo/preview pages are an expected FAIL: record them as FAIL with a "by design — real capture wires at launch" note. Flag, don't "fix," and don't count them against a launch gate unless they ARE the launch surface.

Phase 2 — Fire test submissions

  1. Pre-check form detection. On Netlify: netlify api listSiteForms --data '{"site_id":"<YOUR-SITE-ID>"}' — an empty array plus POST→404 usually means HTML form processing is disabled at the platform level. That alone is a chain-step FAIL; fixing it means a config change + redeploy, which needs the founder's green-light.
  2. Submit like a real user: headless browser (Playwright) — fill, pass client-side validation, click submit, and record all network requests during submit (Phase 3 step 4 needs them). Fallback: curl -X POST with the form fields urlencoded — but a curl POST can't prove client-side validation or tracking fired, so those cells become COULDN'T-VERIFY, not PASS.
  3. Quiz funnels: complete every step through the thank-you screen — and run skippable steps both skipped and filled. Partial-step abandonment is where quiz capture silently dies.
  4. tel:/mailto: can't be fired programmatically. Verify the number/address matches the client's record and is consistent across all pages; the actual ring/delivery is COULDN'T-VERIFY unless the founder places a live call — offer that step.
  5. Booking calendars: book a real test slot ONLY with explicit go-ahead (it lands on the client's calendar); cancel it in cleanup. Chat widgets: send one marked test message; where it routes is checked in Phase 3.

Phase 3 — Verify the chain, per entry point

  1. Platform record: list the form's submissions via the platform API (Netlify: netlify api listFormSubmissions --data '{"form_id":"<FORM-ID>"}'; WordPress: form-plugin entries via wp-cli/REST) — find the entry carrying the TEST marker and capture its ID for cleanup. Empty list right after submitting = wait and re-poll before judging (see the inconclusive-read corollary above).
  2. Notification delivery: confirm a notification hook/recipient is configured (Netlify: netlify api listHooksBySiteId --data '{"site_id":"<YOUR-SITE-ID>"}'). No notification configured = FAIL — a submission nobody is emailed about is a dead lead. If the target is your inbox, confirm the email actually arrived. If it's the client's inbox, ask the founder (or the client on the next call) to confirm receipt — until then it is COULDN'T-VERIFY, not PASS.
  3. <CRM>/automation record: ONLY for chains that claim one exists. Verify the record in the actual system — never trust the automation's own "success" response; some CRM endpoints return success while silently discarding fields, so read the created record and check the field CONTENT, not just its existence. No automation claimed → mark N/A; don't invent a step.
  4. Tracking fires on submit: static grep first (analytics tag, ads pixel) — presence in source ≠ firing (a CSP can silently block a tag that greps clean). Proof is the Phase 2 network capture: a request to the analytics collect endpoint with the lead/conversion event, and to the ads-pixel endpoint with the configured event, during submission. No tag installed → FAIL only if the setup claims tracking; otherwise note "no tracking installed."

Phase 4 — Verification matrix

Save to clients/<slug>/deliverables/qa/lead-flow-verification-YYYY-MM-DD.md (mkdir -p first). One row per entry point; columns: Submitted OK · Platform record · Notification arrived · <CRM> record · Analytics event · Ads-pixel event · Cleanup done. Cells: PASS / FAIL / COULDN'T-VERIFY / N/A — every FAIL and COULDN'T-VERIFY gets a one-line reason plus what would resolve it. Header states the site, live URL, date, and the payload marker used. Close with a blunt verdict: "lead capture is / is not launch-ready" — never "shipped" off an HTTP 200.

Phase 5 — Cleanup

  • Delete each test submission via the platform API, then re-list to confirm it's gone — verify the delete too.
  • Cancel test bookings; delete or flag any <CRM> test records the automation created; note in the matrix anything that couldn't be cleaned so the client can ignore it.
  • If a notification landed in the client's inbox, have the founder flag it as a test.

Write-back

Log an activity note on the client's workspace record (date, site, matrix verdict, path to the matrix file) and update the matching <TASK SYSTEM> task before closing out.

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.