New  ·  Model Context Protocol  ·  Available on every plan

Your AI assistant, plugged into your competitors

Newsletrix runs a built-in MCP server. Connect ChatGPT, Claude, or Cursor and your AI assistant can read your tracked newsletters, brand stats, and recommendations on demand. No exports. No copy-paste. No hallucinated numbers.

Works with Claude, ChatGPT & Cursor Bearer-token auth, revocable anytime Free plan included
What is MCP?

One protocol. Every assistant. Your data.

The Model Context Protocol is an open standard for letting AI assistants talk to external tools and data. Newsletrix implements it so the assistant you already use can answer questions about your real newsletter analytics, not a generic web search.

Ask in plain English

"Summarize last week's competitor sends." "Which brands are running discount campaigns?" "Show me the highest-scoring subject lines." Newsletrix turns the request into structured queries against your account.

Grounded in your data

The assistant only sees what you've tracked. Numbers come from your dashboard, not from training data. If a fact is in Newsletrix, it can quote it. If it isn't, it won't make one up.

Stays in the tool you use

No second dashboard to learn. Live inside Claude Desktop, the ChatGPT connectors panel, or Cursor's chat. Newsletrix is just another capability your assistant can reach for.

You stay in control

Each connection uses a scoped bearer token you generate in settings. Set an expiry, name it per device, revoke it when you're done. Nothing happens behind your back.

Tools

What your assistant can call

Seven structured tools the AI can invoke automatically. No prompt engineering required.

Tool What it returns Use case
list_newsletters Paginated list of newsletters in your account, with subject, brand, and effectiveness score. Browse
get_newsletter Full analysis for a specific send: SWOT, recommendations, scores, tone, compliance. Detail
search_newsletters Search by subject line or brand name across your tracked sends. Search
get_brands Every brand you track with aggregate metrics: avg open, CTR, effectiveness, last send. Aggregate
get_dashboard_stats High-level account stats: total sends, average scores, first / last newsletter, brand count. Aggregate
get_recommendations AI improvement recommendations across your newsletters with rationale and expected impact. Strategy
get_topics Top topics and keywords across the newsletters you track, ranked by frequency. Insights
Example prompts

Things you'll actually ask

Drop these into Claude, ChatGPT, or Cursor once Newsletrix is connected.

"Catch me up"

Summarize every newsletter that landed in my Newsletrix account this week. Group by brand, flag anything with a strong promo or unusual send time.

"Pick a fight"

Compare Acme and Vega over the last 30 days. Where is each one stronger - subject lines, CTAs, cadence? Give me a single paragraph I can share with the team.

"Find the gap"

Which topics show up in less than 10% of competitor sends but get the highest estimated open rates? Sort by topic, return five.

"Steal something"

Show me the three highest-scoring subject lines across every brand I track this quarter. For each one, explain why it likely worked.

"Plan next week"

Pull the top recommendations from Newsletrix and turn them into a checklist I can hand to a junior copywriter. Order by expected business impact.

"Audit a send"

Open the analysis for newsletter UUID abc123. Walk me through the SWOT, then suggest two A/B tests I should run on next week's send.

How it works

Three steps. About two minutes.

Generate a token, paste a config, ask a question. That's the whole setup.

1

Generate an MCP token

Open Settings → AI assistant (MCP) in your Newsletrix dashboard. Click "New token", name it per device, pick an expiry, copy the value.

2

Add Newsletrix to your AI tool

Paste the snippet for Claude Desktop, ChatGPT connectors, or Cursor below. The endpoint is https://beacon.newsletrix.com/mcp.

3

Ask anything

The assistant auto-discovers the tools and calls them when relevant. No special syntax, no chained commands - you just ask.

Setup snippets

Copy, paste, done

One block per client. Replace YOUR_TOKEN with the value from your dashboard.

Claude Desktop

Edit claude_desktop_config.json and add Newsletrix under mcpServers:

{
  "mcpServers": {
    "newsletrix": {
      "type": "http",
      "url": "https://beacon.newsletrix.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}

Restart Claude Desktop. Newsletrix tools appear in the connectors menu.

ChatGPT (Pro / Team / Enterprise)

In ChatGPT settings, open Connectors → Add custom connector and use these values:

Name:           Newsletrix
URL:            https://beacon.newsletrix.com/mcp
Authentication: Bearer
Token:          YOUR_TOKEN

Save and enable. ChatGPT will call Newsletrix tools whenever your prompt looks like newsletter analytics work.

Cursor

Open Settings → MCP in Cursor and paste:

{
  "mcpServers": {
    "newsletrix": {
      "url": "https://beacon.newsletrix.com/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_TOKEN"
      }
    }
  }
}

Cursor will list list_newsletters, get_newsletter, and the rest in the tools panel.

Before & after

The same question, two different worlds

What changes when your assistant can actually read your Newsletrix account.

You ask Without MCP With Newsletrix MCP
"What did my competitors send last week?" Generic guess based on training data Pulls real sends from your account, by brand and date
"Compare Acme vs Vega on subject lines" Hand-roll a CSV first Uses live aggregate metrics per brand
"Top 5 recommendations to ship this week" Best-practice fluff Pulls your actual ranked recommendations
"Show me the analysis for newsletter X" "I don't have access to that" Returns the SWOT, scores, and rationale
End-to-end answer time Re-prompt & copy-paste One question
FAQ

Frequently asked questions

The short version of everything teams ask before flipping it on.

What is MCP?

Model Context Protocol - an open standard that lets AI assistants connect to external tools and data sources in a structured, auditable way. Newsletrix runs an MCP server so any compatible client can read your analytics.

Which assistants are supported?

Claude Desktop, ChatGPT (Pro / Team / Enterprise), Cursor, Zed, and any other MCP-compatible client. Both modern Streamable HTTP and legacy SSE transports are supported.

Is this on every plan?

Yes. Like every Newsletrix feature, the MCP server is included on every plan, including the free tier.

How is access secured?

Bearer-token auth, scoped to your account. You generate, name, and revoke tokens from your dashboard. Tokens never see another user's data.

Can the assistant change my data?

No. The exposed tools are read-only. Your assistant can list, fetch, and search - it can't delete newsletters, edit recommendations, or change settings.

Does it work with self-hosted models?

Anything that speaks MCP can connect. Local LLM frontends like Cursor's local mode, custom Claude SDK agents, and self-hosted clients all work the same way.

Get started

Stop guessing. Start winning.

Join newsletter creators using AI-powered competitor intelligence to ship better content, faster.

No credit card required  ·  Cancel anytime  ·  All features on every plan