Monday morning, and leadership wants to know where the brand stands. To answer, you open the dashboard, jump between domains, social media and takedowns, copy numbers into an email and reword it three times. Half an hour is gone, even though every single number was already in nebty.
Many teams already work in Claude or ChatGPT all day. With nebty MCP, your Brand Protection lives in the same window, and in Claude you will find nebty in the official connector directory. You ask the question the way you would ask a colleague. The assistant pulls the matching findings from your nebty account and answers with domains, profiles and case status that you can check in the dashboard.
What it looks like in the chat
One question is enough. The assistant decides which data it needs, and at every step you can see which nebty tool it calls.
Your assistant
nebty connected
What happened with Samplebrand over the last seven days? Give me a short summary for leadership.
- list_brands
- get_brand_risk_timeline
- list_brand_domains
- list_social_media_scans
- list_takedowns
Since last Monday, three new lookalike domains have appeared, two of them high risk:
| Domain | Risk | Finding |
|---|---|---|
| samplebrand-outlet.shop | High | online, copies your logo |
| samplebrand-eu.com | High | online, login form |
| samplbrand.com | Medium | offline |
On Instagram, two new profiles are waiting for your review. One uses your logo and links to samplebrand-outlet.shop. Of the four running takedowns, one is done: the registrar suspended samplebrand-sale.net.
My suggestion: review samplebrand-outlet.shop and the linked profile first. They appear to belong to the same campaign.
Because the answer is already in the chat, your assistant can keep working with it. It drafts the email to the board or builds a table for the next audit.
What is behind nebty MCP
MCP stands for Model Context Protocol, an open standard that lets AI assistants use data and tools from other applications. Without it, Claude answers from its general knowledge. With nebty MCP, it answers from your monitoring.
47
tools in the connector
41
of them read only
6
set your review status
The read tools cover everything nebty watches for you: lookalike domains with scan evidence, fake profiles on social networks, ads from the Meta Ad Library, fake employees on LinkedIn and your takedowns with their history. The six write tools handle a job that usually takes a lot of clicks. They mark a profile or ad as trusted, or dismiss a finding when it belongs to you or a partner.
Four jobs your team can do in the chat from today
The dashboard stays the place for daily work. The chat is faster as soon as a question touches several areas, or when someone needs the answer who never signs in to nebty.
Leadership
The weekly picture without a slide deck
The assistant reads your key figures, the risk trend of the last few weeks and the state of your takedowns. It turns them into a summary in your own words. If a number looks odd, just ask a follow-up.
"How has the risk for our brand developed over the last 30 days? How many takedowns closed successfully?"
Security and fraud
Prioritise domain findings
The assistant filters by risk, visual match and reachability, then pulls the evidence and scan history for each domain. It explains which signals point to impersonation and what is still unclear.
"Which high-risk domains are online right now and look like our website? Explain the order."
Marketing and social media
Review fake profiles and ads
Social media scans, ads from the Meta Ad Library and the LinkedIn Employee Radar sit in the same conversation. If a profile belongs to a partner, the assistant marks it as trusted once you approve.
"Which active Meta ads use our name and do not come from our own pages?"
Legal and compliance
Takedown status with evidence
For every case the assistant reads the status, the event history, the abuse contacts nebty reached out to and the blocklist submissions. You answer questions from counsel or auditors without an export.
"Show me the history of the takedown for samplebrand-sale.net. Who did nebty contact, and when did they reply?"
The assistant prepares, your team decides
An assistant with access to security data needs clear limits. nebty MCP applies the same permissions as the dashboard. The assistant only sees the brands and records your account is allowed to access.
- Sign-in through OAuth. You sign in on the nebty page. Never enter a password, API key or access token into a chat.
- Read access first. A new connection requests read access. To change a review status, you grant an additional write permission and need the right to edit the brand.
- Takedowns stay in the dashboard. You start new scans, screenshot captures, takedowns and email reports in nebty. The assistant can summarise evidence and suggest a next step. Check the evidence before you act.
- Third-party text is evidence. Fake sites and ads contain text written by fraudsters. The tools tell the assistant that results can include such text. It is evidence, not permission to act.
- Revocable at any time. Under Settings in the nebty dashboard you see the client, the granted permissions and the activity of every connection. You revoke a connection there too.
Connected in three steps
You need a nebty account with access to at least one brand and a suitable plan. Your Claude or ChatGPT plan and your organisation settings must allow custom connectors.
-
Add nebty to your assistant
Claude is the quickest: open nebty in the official Claude Directory and click Connect. You do not need a URL for that.
In ChatGPT, open Plugins and add a custom MCP server. Other clients with remote MCP work the same way. The server URL:
https://mcp.nebty-id.com/mcpReplace older SSE configurations with this /mcp URL.
-
Sign in to nebty
The assistant sends you to the nebty sign-in page at app.nebty.de. Sign in with a six-digit email code, your password or Google, or use the email link instead. If your account uses a second factor, confirm it.
-
Allow access and start asking
Check the client name and the requested permissions, allow access and return to your assistant. A good first prompt: "List my brands and summarise the domain findings for one of them."
Sign-in rollout
The email-code sign-in and signup during connection will be available after the next rollout step. Until then, use your existing nebty account. You can already register through the dashboard.
No nebty account yet?
Choose the create-account option on the nebty sign-in page. You verify your email address with a six-digit code, enter your company and brand details and review the plan, terms and power of attorney. The trial lasts 14 days and requires a payment method, which Stripe Checkout shows before you order. It becomes a paid subscription unless you cancel.
After checkout, you confirm the connection to your assistant separately, because the order alone gives it no access. If the connection request has expired in the meantime, start it again in the assistant without ordering the subscription again. For a new brand, expect monitoring to take some time before results come in.
Questions and help
Which assistants can I connect?
Claude, ChatGPT and any other client that supports remote MCP with OAuth. Your assistant plan and your organisation settings must allow custom connectors. In Claude, nebty is listed in the official connector directory. In ChatGPT and other clients, you add nebty as a custom MCP server.
What do I need on the nebty side?
A nebty account with access to at least one brand and a suitable plan. The assistant sees exactly the brands and areas your account sees in the dashboard.
Can the assistant start takedowns or trigger new scans?
No. You start new scans, screenshot captures, new takedowns and email reports in the nebty dashboard. The assistant reads existing findings and cases, summarises the evidence and suggests a next step. The only thing it can change is the review status of LinkedIn, social media and Meta ad findings, and only with your write permission.
Why is a list empty?
Check the selected brand, the filters and the monitoring status. A new brand often has no findings yet because monitoring needs time to collect results. An empty list only means that no matching records came back. It does not rule out abuse. For large lists, ask for smaller pages and request further pages explicitly, because a partial result is not a complete inventory.
The connection expired or access is denied. What now?
If the sign-in expired or the connection was revoked, start a new connection in the assistant. If an action needs write access, only allow it when you actually want that change. A denial can also mean that your account or the brand's plan does not allow the action. If your organisation blocks custom connectors or the server is unreachable, check your platform settings and try again later.
What happens when I revoke a connection?
Revoking it in the nebty dashboard blocks new requests and token refreshes for that connection. It does not undo earlier changes, stop work already in progress or delete data the assistant provider has already received. You can also disconnect it in the assistant.
What data does my assistant provider receive?
The results of the tools your assistant calls, which is a reviewed subset of your dashboard data. The section on AI assistants in our privacy policy explains how the connection processes data. You will find the link right below these questions.
Connection not working?
Send us the platform, the approximate time and the error message. Please do not include passwords, tokens or sign-in links.
Claude and ChatGPT are trademarks of their respective owners. The listing in the Claude Directory does not mean a partnership with Anthropic or OpenAI.
