v1.0 — live on mcpserver.castersapp.com

Plug your AI into Finland's creator network

An open Model Context Protocol server for CastersApp. Ask about campaigns.popular areas.live traffic.creator reach. Answers come from live records, not a stale export.

Works with Claude Desktop · Claude Code · Cursor · any MCP client

Your AI clientClaude · Cursor
ask() answer
Casters MCPJSON-RPC
MCP servertools + auth
search_campaigns rows
Casters datacampaigns · creators
  1. You ask a question
  2. Claude picks a tool
  3. Server queries live data
  4. Grounded answer
Setup

Connected in under a minute

Drop the server into your client's MCP config. No SDK, no build step, no local database.

{
  "mcpServers": {
    "casters": {
      "command": "npx",
      "args": ["-y", "@casters/mcp-server"],
      "env": { "CASTERS_API_KEY": "sk_live_..." }
    }
  }
}
In action

What a real exchange
looks like

One question, one tool call, an answer built from live rows.

claude — casters mcp
you
tool top_campaigns({ days: 30, limit: 3 })
rows
Kesän hauskimmat tuliaiset4,182 views
Weekend Festival 20263,067 views
Syksyllä uusi harrastus2,415 views
claude Three campaigns lead the last 30 days, with Kesän hauskimmat tuliaiset well ahead at 4,182 views and a 6.1% apply rate — the strongest conversion of the three.
Capabilities

Everything the platform knows,
exposed as tools

Typed, documented and scoped to the caller's own permissions.

Campaign discovery

Search live campaigns by category, city, budget, creator type and timing — the same index that powers the public campaign board, without scraping a single page.

Creator intelligence

Pull creator profiles with follower counts, categories, cities and portfolio links across Instagram, TikTok, YouTube and Facebook — normalised into one shape.

Application flow

List, review, accept or reject applications programmatically — with the same guardrails the dashboard enforces.

Analytics

Reach, impressions, engagement and conversion per campaign or creator, aggregated on the server so your context stays small.

Scoped by key

Every call is bound to an API key. A brand key never sees another brand's data — enforced server-side, not by prompt.

How it works

Three steps, no glue code

1

Generate a key

Create a scoped API key from your Casters dashboard. Brand, creator and read-only scopes are issued separately.

2

Register the server

Paste the config above into your MCP client and restart it. The server advertises its tools automatically on handshake.

3

Just ask

“Which active beauty campaigns in Helsinki still have open slots?” Your assistant calls the right tool and answers from live data.

0Creators reachable
0Finnish cities
0Tools exposed
0Uptime target
FAQ

Questions

The Model Context Protocol is an open standard for connecting AI assistants to external systems. Instead of pasting data into a chat, the assistant calls typed tools directly — so answers come from live records rather than a stale copy.

Every request is authenticated with a scoped API key and authorised server-side against the same rules as the web dashboard. Keys are revocable at any time, and write operations are opt-in per key.

Anything that speaks MCP — Claude Desktop, Claude Code, Cursor, and custom clients built on the MCP SDKs. Both stdio and HTTP/SSE transports are available.

The server is included with your Casters account. Rate limits apply per key and scale with your plan.

Give your assistant
real context

Generate a key and connect in a minute. No migration, no SDK, no lock-in.