Use Nexin in Claude and ChatGPT With One Connector (MCP)

Nexin now works as a connector. You paste one URL into Claude or ChatGPT, sign in with your Nexin email, and every module you own is available in that chat as a tool. Nothing to upload, no code execution to turn on, and when we update a job your AI gets the new version the next time it uses it. The connector is the only way Nexin ships: there are no files to download.

What is the Nexin connector?

The connector is an MCP server. MCP (Model Context Protocol) is the open standard AI apps use to plug into outside tools, and Claude, ChatGPT, Cursor and VS Code all support it. Nexin’s server lives at https://nexin.si/mcp. Every module is one tool in your AI, and each tool runs jobs:

Tool	What your AI uses it for
threads, instagram, x, youtube, pinterest, blog, product, ads, copy, core, nexus	One per module. Your AI picks the module and the job (“post-batch”, “sales-page”, “carousel”) and gets that job’s method, plus its reference files only when a step needs them
brand	Saves your brand files (brand, speaker, proof, funnel) and reads them back in any new chat
prompt	Finds and opens a ready-made prompt from your library, fills the fields it can from your chat and brand files, asks once for the rest, then runs it
check	Counts characters the way each platform does and checks limits, hashtags, links and AI-writing tells
account	Shows which modules you own and what each one holds

You don’t call these yourself. You ask for content as usual and your AI decides when to use them.

Why a connector and not files to upload?

Agent Skills are usually files you upload or copy into a folder. Nexin skips that step for five reasons.

• It works in ChatGPT. ChatGPT has no skills format. Through the connector, ChatGPT runs the same tools and jobs Claude does.

• Nothing to install. Skill files in Claude.ai mean uploading one zip per skill and turning on code execution. The connector is one URL.

• Always current. Updates to every job reach you on their own, with nothing to download again.

• Exact counts. Our jobs ask for real character counts, not guesses. Without code execution a model can only estimate. The check tool counts exactly, including X’s rule that every link counts as 23 characters, and flags anything over a platform’s limit before you post.

• It remembers your brand. The files Core writes (nexin-brand.md, your speaker, proof and funnel files) are saved to your Nexin account, encrypted. A new chat in Claude or ChatGPT reads them on its own, so you never paste your brand again.

How to connect Nexin to Claude

Claude on the web and the desktop app sign in to Nexin for you, so you don’t need a key.

1. Open Settings → Connectors and choose Add custom connector.
2. Name it Nexin and paste https://nexin.si/mcp as the URL.
3. Click Connect, sign in with the email you used at checkout and choose Allow.

Start a new chat and ask: “Use Nexin to check my account.” Claude should list the modules you own.

How to connect Nexin to ChatGPT

1. Open Plugins in the sidebar and click + in the top right to add a custom MCP server. In older versions it’s Settings → Apps & Connectors (turn on Developer mode under Advanced if your plan asks for it), then Create.
2. Name it Nexin, leave Connection on Server URL, paste https://nexin.si/mcp and keep Authentication on OAuth.
3. Sign in with your Nexin email and choose Allow. In a chat, pick Nexin from the tools menu.

Connector options in ChatGPT depend on your plan and workspace settings. If you don’t see them, a workspace admin may need to allow custom connectors.

How to connect Nexin to Claude Code, Cursor and VS Code

These apps use a personal API key instead of a sign-in. Create one on the Setup page in your Nexin library (one key per computer, so you can revoke one without touching the others), then:

claude mcp add --transport http nexin https://nexin.si/mcp --header "Authorization: Bearer YOUR_KEY"

For Cursor (~/.cursor/mcp.json) or VS Code (.vscode/mcp.json):

{ "mcpServers": { "nexin": { "url": "https://nexin.si/mcp", "headers": { "Authorization": "Bearer YOUR_KEY" } } } }

The Setup page fills your key into both snippets when you create it.

What a request looks like

Say you own Threads and ask for “a week of Threads posts about meal prep”. Your AI:

1. Calls the threads tool with the job post-batch, which returns that job’s method.
2. The first step needs your brand file, so it reads it with brand, or offers to run the Core interview if you haven’t saved one yet.
3. Writes eight posts in the shapes the job calls for.
4. Runs them through check: counts, the 500-character limit, how many are under 120 characters, links in the post body, em dashes and repeated openers. It fixes what fails before you see the batch.

Every user sees the same tools. If you ask for something in a module you don’t own, that module’s tool answers that it isn’t on your account, with the link to get it. Your AI won’t try to imitate a module you don’t have.

After you buy another module

Nothing to install. Every module is already a tool in your AI, and the server checks what you own on every call, so the new module works on your next request. If your AI still says you don’t own it, start a new chat. Still stuck? Refresh the connector once:

• Claude: Settings → Connectors → Nexin → the ⋮ menu → Refresh tools list. Then start a new chat.
• ChatGPT: Plugins → Nexin under Installed → refresh or reconnect (older versions: Settings → Apps & Connectors → Nexin → Refresh). Then start a new chat.
• Claude Code: run /mcp, choose nexin and Reconnect, or start a new session.
• Cursor and VS Code: turn the Nexin server off and on in its MCP settings.

If something still looks out of date, disconnect Nexin and connect it again. After a purchase, your library shows these same steps.

FAQ

Is the connector included with my purchase? Yes. Every Nexin module works through the connector at no extra cost, and it serves exactly the modules you own.

Does Nexin see my posts? Your drafts pass through the check tool when your AI checks them, and they aren’t stored: our usage log keeps only sizes and which tool and job were used. The one thing we keep is your brand files, when your AI saves them: encrypted, readable only by you and the apps you connect, and you can download or delete them from Brand files in your library.

Can I use it for more than one brand? Yes. Ask your AI to save files under a brand name (for example “client-two”) and to use that brand in later chats. You can keep up to 50.

Do I need a paid Claude or ChatGPT plan? Custom connectors and longer outputs generally need a paid plan. The exact rules depend on the app and your plan.

Can I disconnect it? Yes. Your Setup page lists every connected app and every key, and you can disconnect or revoke each one.