Instagram MCP Server: Connect AI Agents to Instagram Data

If you are building an AI agent that needs Instagram data, MCP is the new way to do it. Instead of writing a scraper, hosting it, then teaching your agent how to call it, you point any MCP-compatible client at an Instagram MCP server and the agent has typed tools available immediately. This guide explains what an Instagram MCP server is, what tools it should expose, and how to set up CreatorCrawl's hosted one in the major MCP clients.

#What an Instagram MCP server does

MCP (Model Context Protocol) is a standard from Anthropic that lets AI agents discover and call tools on a server through a structured protocol. An Instagram MCP server exposes Instagram data as those tools. The agent reads the tool list, knows what arguments each one takes, and can call them in any combination.

For Instagram specifically the useful tools are the ones that let an agent answer real questions:

Tool	What an agent uses it for
get_instagram_profile	"How many followers does @zuck have right now"
get_instagram_basic_profile	"Quick stats for this handle, no posts"
get_instagram_posts	"Pull the last 30 posts from this brand and analyze captions"
get_instagram_post	"Pull a single post by URL with full metadata"
get_instagram_comments	"Summarize the sentiment in the comments on this post"
get_instagram_reels	"Which reels are performing best for this creator"
search_instagram_reels	"Find recent reels under a search term"
get_instagram_transcript	"Get the spoken transcript of this reel"
get_instagram_story_highlights	"List the highlight reels on this profile"
get_instagram_highlight_details	"Pull stories inside a specific highlight"
get_instagram_embed	"Get the embeddable HTML for a post"

CreatorCrawl's MCP server exposes all of these, plus more, at a single endpoint.

#Why MCP instead of raw API calls

A regular REST API works fine if you are writing the orchestration yourself. The agent benefits start once you want the agent to reason about which calls to make.

With MCP the agent discovers the tools, knows the parameter types, and chains calls without orchestration code. You write the prompt, the agent picks the tools, and you get structured output. The same MCP server works across Claude Desktop, Cursor, Windsurf, custom agents built on the OpenAI Responses API, and anything else that speaks the protocol.

You also avoid the per-site adapter problem. If your agent needs Instagram and TikTok and YouTube data, the MCP version is one config line per source. The REST version is three integrations, three auth flows, three sets of error handling.

#Setup

#Prerequisites

1. Sign up for CreatorCrawl and grab an API key from the dashboard.
2. Pick an MCP client: Claude Desktop, Cursor, Windsurf, or your own.

The endpoint is https://app.creatorcrawl.com/api/mcp. Same URL, all platforms.

#Claude Desktop

Open Claude Desktop settings, go to MCP Servers, paste this config:

{
  "mcpServers": {
    "creatorcrawl": {
      "url": "https://app.creatorcrawl.com/api/mcp",
      "headers": {
        "x-api-key": "your_api_key_here"
      }
    }
  }
}

Restart Claude Desktop. The Instagram tools appear in the tools panel alongside the other CreatorCrawl tools.

#Cursor

Cursor settings, MCP Servers, same config as above. Save and the tools register with the agent panel.

#Windsurf

Windsurf settings, MCP, paste the same JSON, restart. Cascade picks up the tools automatically.

#Custom client

If you are building your own agent and want to call CreatorCrawl directly:

import { Client } from '@modelcontextprotocol/sdk/client/index.js'
import { StreamableHTTPClientTransport } from '@modelcontextprotocol/sdk/client/streamableHttp.js'

const transport = new StreamableHTTPClientTransport(
  new URL('https://app.creatorcrawl.com/api/mcp'),
  {
    requestInit: {
      headers: { 'x-api-key': 'your_api_key_here' },
    },
  },
)

const client = new Client({ name: 'my-agent', version: '1.0.0' })
await client.connect(transport)

const profile = await client.callTool({
  name: 'get_instagram_profile',
  arguments: { handle: 'zuck' },
})

#Example agent prompts

Once connected, here are prompts that exercise the Instagram tools and combine them with reasoning. The agent picks which tool to call.

#Find creators in a niche

"Find 10 Instagram creators in the home cooking niche with between 50k and 500k followers. Pull each one's bio and last 5 posts. Rank them by engagement rate."

The agent searches users, fetches profiles, pulls recent posts, calculates engagement, returns a ranked list.

#Competitor monitoring

"Pull the last 20 posts from @nike on Instagram and @adidas. Compare average likes, average comments, and posting frequency. Which brand is performing better right now."

Two profile calls, two post-history calls, agent does the math.

#Hashtag analysis

"What are people posting under #ifyk this week. Pull the top 50 posts and summarize the themes."

Hashtag-posts call plus an analysis pass on the captions.

#Comment sentiment

"Get the comments on this Instagram post [URL] and tell me what fans love and what they are complaining about."

One tool call, the model handles the synthesis.

#Influencer due diligence

"Give me a profile audit of @creatorhandle. Check their bio links, recent post performance, engagement rate trend, and whether their comments look organic or botted."

The agent chains profile, posts, and comments calls, then reasons about the patterns.

#What costs what

Each tool call uses 1 credit. Credit packs:

Pack	Credits	Price
Free	250	$0
Starter	5,000	$29
Pro	20,000	$99
Scale	100,000	$299

Credits never expire. No subscription. You pay per call, agents are notoriously hard to budget for, this pricing reflects that.

#Common questions

Is this the official Instagram API. No. The official Instagram Graph API is gated and limited to your own pages. CreatorCrawl returns the same data your browser would see for any public profile, just structured and stable.

Does it work for private accounts. No. Public data only.

What rate limits apply. None per second. The only ceiling is your credit balance.

Can I use this from non-Anthropic agents. Yes. MCP is an open protocol. OpenAI Responses API, custom LangChain agents, and any other MCP-compatible runtime all work.

#Next

If you are setting up MCP for the first time, start with the main MCP docs. If you also need TikTok, YouTube, LinkedIn, or Reddit data, the same endpoint covers all of them. Browse the full integration list or sign up for free to start.