Auto-approve Skool members with n8n + GPT-4o AI screening

End-to-end recipe: when someone applies to your Skool community, an n8n workflow runs a quality filter (LinkedIn check + GPT-4o AI screening of their application answer) and auto-approves the ones that pass — no manual review required.

🚀 Ready-to-import template available on n8n.io

→ Import the workflow on n8n.io

The template includes the full Apify HTTP Request setup, GPT-4o screening node with prompt, error handling, and Telegram audit trail. Import it into your n8n instance and just plug in your credentials.

This is the exact pattern running in production for Cágala, Aprende, Repite — handles ~3 applicants/day, approves ~70% automatically, queues the rest for human review.

Below: the full walkthrough if you want to understand how it works (or build it from scratch in another tool).

What you’ll build

                                         ┌─────────────────────┐
[Skool] ──→ pending applicant ──→        │ n8n workflow        │
                                         │                     │
                                  ┌──→   │ 1. Schedule trigger │
                                  │      │    (every 15 min)   │
                                  │      │                     │
                                  │      │ 2. members:pending  │   ──→ Apify actor
                                  │      │    (read pending)   │
                                  │      │                     │
                                  │      │ 3. Filter           │
                                  │      │    (LinkedIn? bio?  │
                                  │      │     country? etc.)  │
                                  │      │                     │
                                  │      │ 4. members:approve  │   ──→ Apify actor
                                  │      │    (for each pass)  │
                                  │      │                     │
                                  │      │ 5. Notify Telegram  │
                                  │      │    (audit trail)    │
                                  │      └─────────────────────┘
                                  │
                                  └──── repeats every 15 min

Prerequisites

Step 1: Get cookies once

Add a manual workflow that runs auth:login and saves the cookies. Run this every 3 days (or whenever WAF_EXPIRED triggers your error handler).

HTTP Request node — Login:

Set node — Save cookies to data store:

Step 2: Main workflow — auto-approve

Schedule this every 15 minutes.

Node 2.1 — Schedule Trigger

Interval: every 15 minutes.

Node 2.2 — Read cookies from store

(adapt to your store)

Node 2.3 — HTTP Request — members:pending

The response (in dataset[0].members) is an array of pending applicants:

{
  "success": true,
  "members": [
    {
      "id": "request-id-32hex",
      "memberId": "user-id-32hex",
      "email": "applicant@example.com",
      "firstName": "Jane",
      "lastName": "Smith",
      "bio": "Founder, Buenos Aires, building a SaaS for vets",
      "linkedinUrl": "https://www.linkedin.com/in/janesmith",
      "country": "AR",
      "appliedAt": "2026-05-06T18:23:00Z",
      "applicationAnswer": "I want to learn how to fundraise"
    }
  ]
}

Node 2.4 — Check error

IF node: ` === false`

Node 2.5 — Split out members

Split the members array into individual items.

Node 2.6 — Filter — your quality criteria

This is where you encode your community standards. Examples:

// In an n8n Function or Filter node:
const m = $json;

// Required: LinkedIn URL present
if (!m.linkedinUrl) return false;

// Required: bio at least 30 chars
if (!m.bio || m.bio.length < 30) return false;

// Required: country in your target market
const allowedCountries = ['AR', 'CL', 'CO', 'MX', 'PE', 'ES'];
if (!allowedCountries.includes(m.country)) return false;

// Reject if applicationAnswer is empty or just spam
if (!m.applicationAnswer || m.applicationAnswer.length < 20) return false;

return true;  // pass

For more sophisticated filtering, you can call:

Node 2.7 — HTTP Request — members:approve

For each member that passes:

Note: use memberId, not id. The id field on the pending request is the request itself; memberId is the canonical user ID Skool needs for the approve call.

Node 2.8 — Notify Telegram (audit trail)

Send yourself a message: “Auto-approved: Jane Smith (LinkedIn: …). Application: …”

Optional: also send to a channel where the rest of your admin team can audit decisions.

Step 3: Members that DON’T pass

Don’t auto-reject — manual review is safer. Three options:

  1. Leave in pending — they stay in Skool’s queue, you review when you have time
  2. Notify you on Telegram with their info → you click /approve or /reject reply commands
  3. Send to NocoDB / Airtable with a “queue” status — review from your dashboard

The pattern at CAR uses option 2: a Telegram bot sends each non-auto-approved applicant with /approve {memberId} and /reject {memberId} quick actions.

Production gotchas

Rate limits

members:approve is a write action. Skool limits writes to ~25/min. If you have a sudden burst (e.g. you launched on Product Hunt), the actor’s rate limiter blocks calls — they queue, not fail. If you go over Skool’s hard limit you’ll see RATE_LIMIT errors; back off 60s and retry.

Idempotency

If your workflow runs twice for the same applicant (n8n retry, manual run, etc.), the second members:approve returns SKOOL_API_ERROR “cannot update to same role”. This is a no-op — they’re already approved. Filter by current role before calling approve, or treat the duplicate error as success.

What if members:pending returns no one?

Empty list (members.length === 0) is normal — no applicants in queue. The success field is still true. Don’t alert on empty.

Cookies expiring mid-run

If your scheduled run hits WAF_EXPIRED mid-loop, only some approvals will succeed. The error handler should:

  1. Trigger the login workflow → save new cookies
  2. Re-run only the unprocessed members (not the whole members:pending again — that returns the new pending list which excludes the ones you already approved)

Where to go next

More in Recipes