How to Connect Claude to Live Social Trends via MCP

Connecting Claude to live data via MCP lets the model query real-time external sources instead of relying only on training data or pasted text. Model Context Protocol (MCP) is an open-source standard created by Anthropic for linking AI applications to tools, data sources, and workflows. Claude supports remote MCP servers as custom connectors on Free, Pro, Max, Team, and Enterprise plans; Free users are limited to one custom connector.

Remote MCP servers run on the public internet and communicate with Claude from Anthropic’s cloud infrastructure. Once added, Claude can call the tools exposed by the server during a conversation. This guide shows how to attach a live social-trend MCP—using TrendsAGI’s hosted endpoint or a local stdio server—so Claude can retrieve velocity, AI insights, analytics, recommendations, and crisis events without leaving the chat.

What MCP gives Claude

According to modelcontextprotocol.io, MCP is an open standard that lets AI applications connect to external data sources, tools, and workflows. Claude treats a remote MCP server as a custom connector. Users add the server URL under Customize > Connectors (or Organization settings for Team/Enterprise owners), as described in Claude.ai Documentation. After connection, Claude can enable or disable the connector per conversation via the + / Connectors menu.

TrendsAGI runs a hosted MCP server at https://api.trendsagi.com/mcp that supports OAuth sign-in for Claude and other MCP clients. Hosted tool calls count toward the account’s API limits; each OAuth connection receives its own key listed under Profile > API Keys. The Developer plan includes 1 API integration connection and 100 API calls per day, with overage billing above the included usage.

Two ways to connect TrendsAGI to Claude

You can use the publicly reachable hosted server (recommended for Claude.ai, Claude Desktop, and Cowork) or run a local stdio server from the Python package (useful for Claude Code and local development).

Hosted remote MCP (OAuth)

Claude Code can register the TrendsAGI remote server with a single command:

claude mcp add --transport http trendsagi https://api.trendsagi.com/mcp

On claude.ai or Claude Desktop, open Customize > Connectors (or the organization connectors page) and paste the same URL as a custom connector. Complete the OAuth flow when prompted. The server must be publicly reachable because Claude connects from Anthropic’s infrastructure.

Local stdio MCP

Install the package with MCP extras (Python 3.10+):

pip install "trendsagi[mcp]"

Set the API key and start the server:

export TRENDSAGI_API_KEY=your_key_here
trendsagi mcp

Or register it with Claude Code:

claude mcp add trendsagi --env TRENDSAGI_API_KEY=your_key_here -- trendsagi mcp

JSON configuration for Claude Desktop, Cursor, or VS Code follows the same pattern:

{
  "mcpServers": {
    "trendsagi": {
      "command": "trendsagi-mcp",
      "env": {
        "TRENDSAGI_API_KEY": "your_key_here"
      }
    }
  }
}

Add --allow-writes (or the corresponding flag when constructing the server) only when write tools are required. By default the MCP tools are read-only.

What Claude can call after connection

TrendsAGI MCP tools expose the same surface as the Python SDK’s TrendsAGITools. Read-only tools available by default include list_trends, list_trend_categories, search_trend_insights, get_trend, get_trend_ai_insights, get_trend_analytics, list_recommendations, list_crisis_events, list_watchlist, get_api_usage, and find_capability (version 0.14.0+). Write tools (add_to_watchlist, update_recommendation) appear only when the server is started with the allow-writes flag.

Velocity is a measured snapshot rate (posts per hour from stored snapshots), not a forecast or unique-user count. Recommendation confidence reflects evidence completeness, not probability of success. Ad-platform connectors preview changes before any explicit execution.

Safety controls for agent builders

Keep the MCP server read-only unless a specific workflow needs writes. Use get_api_usage to inspect remaining quota before long-running agent loops. When the agent proposes an ad change, surface the preview to a human for approval rather than auto-executing. Hosted tool calls consume the same daily allowance as ordinary REST or SDK calls; monitor usage so agents do not exhaust the Developer-plan limit of 100 calls per day.

Python SDK and agent tools (side-by-side with MCP)

The same capability is available outside MCP through the official client:

import os
from trendsagi import TrendsAGIClient, TrendsAGITools

client = TrendsAGIClient(api_key=os.environ["TRENDSAGI_API_KEY"])
tools = TrendsAGITools(client)  # read-only by default
# tools = TrendsAGITools(client, allow_writes=True)  # opt-in writes

# OpenAI-style tool definitions
openai_tools = tools.openai()
# Claude tool-use format
anthropic_tools = tools.anthropic()
# LangChain tools
lc_tools = tools.langchain()

Execute a tool call and unwrap the result:

result = tools.call("list_trends", {"category": "Technology", "sort_by": "velocity", "limit": 5})
if "error" in result:
    print(result["error"], result.get("detail"))
else:
    trends = result["result"]["trends"]
    print(trends[0]["id"], trends[0]["name"], trends[0]["volume"])

End-to-end agent example

A realistic agent flow: discover a rising trend in a category, inspect its analytics and AI insight, then propose a campaign adjustment for human review.

import os
from trendsagi import TrendsAGIClient, TrendsAGITools

client = TrendsAGIClient(api_key=os.environ["TRENDSAGI_API_KEY"])
tools = TrendsAGITools(client)

# 1. Find high-velocity trends in Technology
list_res = tools.call("list_trends", {
    "category": "Technology",
    "sort_by": "velocity",
    "limit": 3
})
trend = list_res["result"]["trends"][0]
trend_id = trend["id"]

# 2. Pull 7-day analytics
analytics = tools.call("get_trend_analytics", {
    "trend_id": trend_id,
    "period": "7d"
})

# 3. Read AI insights (summary, themes, sentiment, marketing angles)
insights = tools.call("get_trend_ai_insights", {"trend_id": trend_id})

# 4. Check remaining quota before further calls
usage = tools.call("get_api_usage", {})

# 5. Surface a human-readable proposal (no automatic ad execution)
print("Proposed action for review:")
print(f"Trend: {trend['name']} (id={trend_id})")
print("Review analytics and insights above, then decide whether to adjust bids or creative.")
print("Usage snapshot:", usage.get("result"))

The agent stops at the proposal. Any subsequent watchlist addition or recommendation update requires the allow-writes flag and an explicit human confirmation step.

Practical tips for Team and Enterprise roll-outs

Owners configure connectors at the organization level. Individual users can still enable or disable a connector inside each conversation. Because remote servers are reachable from Anthropic’s network, review the MCP provider’s authentication (OAuth for TrendsAGI’s hosted endpoint) and data-handling practices before org-wide enablement. Prefer the read-only configuration until write use-cases are audited.

Frequently asked questions

What is MCP and how does Claude use remote MCP servers as custom connectors?

MCP is an open-source standard for connecting AI applications like Claude to external data sources, tools, and workflows. Claude treats a remote MCP server as a custom connector; once added, the model can invoke the tools the server exposes during a conversation.

How do I add a remote MCP server URL in Claude (web, Desktop, or Code)?

On claude.ai or Claude Desktop, enter the URL under Customize > Connectors (or Organization settings for Team/Enterprise). In Claude Code use claude mcp add --transport http <name> <url>. For TrendsAGI the hosted URL is https://api.trendsagi.com/mcp.

What are the plan limits and authentication options for custom connectors?

Custom connectors are available on Free, Pro, Max, Team, and Enterprise plans; Free is limited to one custom connector. TrendsAGI’s hosted server supports OAuth; each connection receives its own API key. Hosted tool calls count against the account’s daily API quota (Developer plan includes 100 calls per day).

How do I connect Claude specifically to a live trend/social-data MCP such as TrendsAGI?

Use the hosted endpoint https://api.trendsagi.com/mcp with OAuth, or install pip install "trendsagi[mcp]" and run trendsagi mcp with TRENDSAGI_API_KEY set. Claude Code one-liner for the hosted server: claude mcp add --transport http trendsagi https://api.trendsagi.com/mcp.

What tools or data can Claude access once connected, and are they read-only?

Claude can list trends (by velocity, growth, volume), fetch AI insights, analytics, recommendations, crisis events, and watchlist items. Tools are read-only by default; write tools require an explicit allow-writes flag when the MCP server starts.

What security and privacy risks should I review before connecting an unverified MCP server?

Remote servers must be publicly reachable and receive traffic from Anthropic’s infrastructure. Confirm the provider’s authentication method, whether tool calls are logged, and that write tools stay disabled until needed. Prefer OAuth-backed hosted endpoints and inspect remaining API quota with get_api_usage before running agents at scale.

How we researched this

Prepared by TrendsAGI Research. Facts checked on 18 October 2026 against the sources listed below.

Capability descriptions and code use TrendsAGI's documented REST API, Python SDK and MCP server. External facts come from the sources below.

Claim	Applies to	Source
MCP is an open-source standard for connecting AI applications like Claude to external data sources, tools, and workflows.	All plans	modelcontextprotocol.io
Custom connectors using remote MCP are available on Claude, Cowork, and Claude Desktop for Free, Pro, Max, Team, and Enterprise plans; Free users are limited to one custom connector.	Free, Pro, Max, Team, Enterprise	support.claude.com
Remote MCP servers communicate with Claude over the internet; Claude connects to them from Anthropic’s cloud infrastructure, so the server must be publicly reachable.	All plans	support.claude.com
Users add a custom connector by entering the remote MCP server URL in Customize > Connectors (or Organization settings for Team/Enterprise owners).	All plans	claude.com
TrendsAGI runs a hosted MCP server at https://api.trendsagi.com/mcp that supports OAuth sign-in for Claude and other MCP clients.	All plans	pypi.org
Claude Code can add the TrendsAGI remote server with: claude mcp add --transport http trendsagi https://api.trendsagi.com/mcp.	All plans	pypi.org
The TrendsAGI Python package ships a local stdio MCP server via pip install "trendsagi[mcp]" and the command trendsagi mcp (or claude mcp add with TRENDSAGI_API_KEY).	All plans	pypi.org
TrendsAGI MCP tools are read-only by default; write tools (add_to_watchlist, update_recommendation) require an explicit allow-writes flag.	All plans	pypi.org
TrendsAGI Developer plan includes 1 API integration connection and 100 API calls per day (overage billing above included usage).	Developer	pypi.org
Hosted TrendsAGI MCP tool calls count toward the account’s API limits; each OAuth connection receives its own key listed under Profile > API Keys.	All plans	pypi.org
Claude can enable or disable configured connectors per conversation via the + / Connectors menu in the chat interface.	All plans	support.claude.com

We re-research and rewrite this page at least every four months; the updated date only changes when the facts are re-checked.

How to Connect Claude to Live Social Trends via MCP

What MCP gives Claude

Two ways to connect TrendsAGI to Claude

Hosted remote MCP (OAuth)

Local stdio MCP

What Claude can call after connection

Safety controls for agent builders

Python SDK and agent tools (side-by-side with MCP)

End-to-end agent example

Practical tips for Team and Enterprise roll-outs

Frequently asked questions

What is MCP and how does Claude use remote MCP servers as custom connectors?

How do I add a remote MCP server URL in Claude (web, Desktop, or Code)?

What are the plan limits and authentication options for custom connectors?

How do I connect Claude specifically to a live trend/social-data MCP such as TrendsAGI?

What tools or data can Claude access once connected, and are they read-only?

What security and privacy risks should I review before connecting an unverified MCP server?

How we researched this

Sources

Try it with TrendsAGI

An MCP Server for Real-Time Trends: Setup Guide

OpenAI Function Calling with Live Trends API (Python)

How to Send Trending Topics to Zapier with Webhooks

Turn trend signals into timely decisions

What MCP gives Claude

Two ways to connect TrendsAGI to Claude

Hosted remote MCP (OAuth)

Local stdio MCP

What Claude can call after connection

Safety controls for agent builders

Python SDK and agent tools (side-by-side with MCP)

End-to-end agent example

Practical tips for Team and Enterprise roll-outs

Frequently asked questions

What is MCP and how does Claude use remote MCP servers as custom connectors?

How do I add a remote MCP server URL in Claude (web, Desktop, or Code)?

What are the plan limits and authentication options for custom connectors?

How do I connect Claude specifically to a live trend/social-data MCP such as TrendsAGI?

What tools or data can Claude access once connected, and are they read-only?

What security and privacy risks should I review before connecting an unverified MCP server?

How we researched this

Sources

Try it with TrendsAGI

Related analysis

An MCP Server for Real-Time Trends: Setup Guide

OpenAI Function Calling with Live Trends API (Python)

How to Send Trending Topics to Zapier with Webhooks

Turn trend signals into timely decisions