LoginSchedule demo
model context protocol

Your advocacy program,
accessible to any LLM.

The EveryoneSocial MCP server lets Claude, Cursor, ChatGPT, or any agent that speaks the Model Context Protocol read and act on your program directly. No screen-scraping. No middleware. The same role-based permissions you already have.

what is mcp

A standard handshake between LLMs and the systems they work with.

The Model Context Protocol is an open spec from Anthropic that lets any LLM call tools, read context, and take action across the apps you actually use — Slack, your CRM, your data warehouse, your advocacy program. Instead of every vendor inventing a different "AI integration," everyone implements the same protocol once.

We shipped an MCP server on top of EveryoneSocial because the question we kept hearing from customers was: "can our internal AI assistant just talk to our advocacy program directly?" Now it can.

what it can do

Four tool families. Composable into anything.

Every tool respects the role and topic permissions of the user who authorized the connection. An employee gets employee scope. An admin gets admin scope. Compliance gates still apply.

es.content.search

Find shareable posts across the program by topic, audience, freshness, or compliance state.

"Show me the three highest-performing AI thought-leadership posts from this month I can share."
es.posts.suggest

Generate a draft post in the program voice from a URL, prompt, or context — tagged to the right topic.

"Draft a Sales-team LinkedIn post about today's product update, in our voice, 3 variations."
es.analytics.read

Pull program metrics — shares, reach, top contributors, topic performance — over any window.

"Which topics drove the most pipeline in the last quarter, and who were the top 5 contributors?"
es.admin.query

List users, segments, topics, and bindings. Read-only by default; mutations gated by your role.

"Which Sales users are missing LinkedIn auth and haven't shared this month?"
connect

Add it to Claude, Cursor, or your own agent in 60 seconds.

Drop the server URL into any MCP-compatible client. OAuth handles the rest — users authorize their own scope, and your admins control which tools are exposed at the program level.

{
  "mcpServers": {
    "everyonesocial": {
      "url": "https://mcp.everyonesocial.com/v1",
      "transport": "http",
      "auth": {
        "type": "oauth",
        "authorize_url": "https://app.everyonesocial.com/oauth/authorize"
      }
    }
  }
}

Snippet shown for Claude Desktop. Cursor, Continue, and custom MCP clients use the same URL. Full reference docs ship with the public beta — request early access →

why it matters
No more screen-scraping.

Your internal AI talks to EveryoneSocial through a typed, audited interface — not a brittle browser automation that breaks every release.

Permissions follow the user.

OAuth-scoped per-user. The agent inherits the same role, topic visibility, and compliance gates the user already has in the app.

Works with anything.

Claude, Cursor, Continue, OpenAI's MCP-compatible clients, your own agents. The protocol is the integration.

Ready to plug your advocacy program into AI?

Early-access customers help shape the tool surface and roadmap.

Request access →Book a demo