What is a Threads MCP server?

A server that implements the Model Context Protocol. It packages Throk's public Threads data (search, trending boards, trends, accounts, ladders and leaderboards) as 18 tools that Claude, Cursor and other AI clients can call directly.

Which clients are supported?

Any client that supports MCP, including Claude Code, Claude Desktop, ChatGPT, Codex and Cursor. Add the MCP URL, then sign in with your Throk account or pass your API key.

Choose an API plan at developer.throk.ai. Plans start at US$30/month and are billed in credits. MCP (with an API key or a Throk account sign-in) and the REST API use the same credits. The web app's Free and Creator plans are billed separately from API plans.

How is MCP different from the REST API?

The data is identical. Use MCP when an AI agent makes the calls; use REST when you call from your own code or service. REST endpoints return larger result sets (semantic search, for example, returns up to 3,000 posts per call).

THREADS MCP SERVER

One command,and your AI can read Threads.

Throk MCP Server gives Claude, Cursor and other AI clients 18 tools for public Threads data. Your agent stops guessing. It searches viral posts, checks trends and reads accounts, and every answer comes back with real numbers.

Get your API key See API and REST docs

# Add to Claude Code with one command
claude mcp add --transport http throk \
  https://premium-api.throk.ai/mcp \
  --header "Authorization: Bearer sk-throk-..."

Works with Claude Code, Claude Desktop, ChatGPT, Codex, Cursor and any other MCP client. You can also skip the key and sign in with your Throk account. Full setup guide

WHAT IS MCP

A standard way for AI to call Threads data directly

MCP (Model Context Protocol) is an open standard for connecting AI clients to external tools. Once you add Throk MCP, your AI stops answering Threads questions from memory. It calls a tool, reads back real data, then answers.

Ask “what topics are rising this week?” and the agent pulls the trending board and keyword trends on its own. Ask “how fast is this account gaining followers?” and it fetches the account's follower history and follower ladder percentile. You don't write any code.

FULL TOOL LIST

18 tools for search, trends and accounts

Same key and credits as the REST API. Every tool works out of the box.

Search and inspiration

search_posts: Keyword search across public posts, with engagement data
semantic_search_posts: Semantic search: describe an idea in one sentence, find similar viral posts
get_inspiration: Inspiration engine: content angles drawn from semantic clusters
get_cool_replies: Top replies leaderboard, for spotting high-engagement reply patterns

Trends and timing

get_hot_posts: Trending board of viral posts, ranked by how fast they gain engagement
get_keyword_trend: Keyword volume over time: is a topic rising or fading?
get_post_like_growth: Viral post leaderboard: the posts gaining likes fastest
get_power_hours: Power hours heatmap: the best times to post

Accounts and rankings

get_user: Account profile and core metrics
get_user_top_threads: An account's top posts over time
get_user_follower_history: Follower count history
get_follower_growth: Follower growth leaderboard
get_follower_ladder: Follower ladder: where an account ranks, as a percentile
get_like_ladder: Like ladder: where a post's performance ranks, as a percentile

Live data and forecasts

forecast_posts: Viral forecast: live views, projected 48-hour views and the odds of hitting a target
get_live_posts: Live views, likes, replies, reposts and images for any post
get_live_users: Live follower count and profile for any account
get_balance: Check your remaining credits (free)

USE CASES

What your agent can do once it's connected

Automate content research

“Find this week's viral posts about AI and give me three angles I could write.” The agent searches, reads the data and comes back with a conclusion.

Account audits and competitor tracking

Pull follower history, ladder rank and top posts for your account and your competitors, and get a comparison report from one conversation.

Build it into your product

MCP and REST share one key. Prove the workflow in Claude first, then use REST to bring the same data into your own service.

What does it cost?

API plans start at US$30/month. MCP and the REST API draw from the same developer credits, and each tool's credit cost is listed in its description.

See API plans and pricing →

Frequently asked questions

What is a Threads MCP server?: A server that implements the Model Context Protocol. It packages Throk's public Threads data (search, trending boards, trends, accounts, ladders and leaderboards) as 18 tools that Claude, Cursor and other AI clients can call directly.
Which clients are supported?: Any client that supports MCP, including Claude Code, Claude Desktop, ChatGPT, Codex and Cursor. Add the MCP URL, then sign in with your Throk account or pass your API key.
How is it priced?: Choose an API plan at developer.throk.ai. Plans start at US$30/month and are billed in credits. MCP (with an API key or a Throk account sign-in) and the REST API use the same credits. The web app's Free and Creator plans are billed separately from API plans.
How is MCP different from the REST API?: The data is identical. Use MCP when an AI agent makes the calls; use REST when you call from your own code or service. REST endpoints return larger result sets (semantic search, for example, returns up to 3,000 posts per call).

Full setup guide

Connect in five minutes. Let your AI check the numbers.

Paste one command and sign in with your Throk account (or create an API key). Your agent now has eyes on Threads.

Get an API key See the full developer docs