For AI assistants
Vet creators, influencers, and athletes from inside Claude, ChatGPT, Cursor, or any Model Context Protocol client — without leaving the conversation. Ask for a creator's 1–100 trust score, the full 7-agent breakdown, or a fair sponsorship rate, and get an answer in the chat.
The MCP server is a thin client over the public CreatorScore API. It holds exactly one secret — your API key — and makes plain HTTPS calls to creatorscore.io/api/v1. It has no database access and no scraping, ML, or provider credentials. Every score is produced server-side by the same gated scoring pipeline brands use on the dashboard.
That means the assistant is strictly less privileged than a logged-in brand: credit metering, rate limits, and the daily fresh-scrape safety cap all apply automatically. An agent can't run up a surprise bill — see Costs & safety below.
| Tool | What it does | Cost |
|---|---|---|
help | Shows the menu of everything CreatorScore can do and the input each option needs — the assistant offers this when you connect. | Free |
check_account | Your credit balance and per-platform pricing. | Free |
score_creator_footprint | From a single handle, discover the creator's accounts across platforms (their bio's link-in-bio + Google-verified matches, each with a confidence) and return the unified whole-footprint score, breakdown, and worst risks. The easiest way to vet a creator. | 1 credit discovery + 1 per scored account |
vet_creator | Vet one creator across all their accounts at once (you supply the handles) — unified cross-platform CreatorScore, per-platform breakdown, and worst risks on any account. Never scrapes. | 1 credit per scored account |
score_creator_fresh | Runs a fresh, full scoring pass (scrape + transcribe + score). Requires explicit confirmation. | 1–3 credits / platform |
check_score_job | Poll a fresh-scoring job until it's done. | Free |
calculate_fair_rate | What to charge or pay for a sponsored post — a fair rate, confidence band, and multiplier breakdown (OCRS). Pure math, no scraping or scoring. | Free — 0 credits |
You need Node.js 18 or newer installed (that's what provides npx). There's nothing else to install or build — npx fetches and runs creatorscore-mcp automatically the first time your assistant starts it.
Edit claude_desktop_config.json (Settings → Developer → Edit Config) and add:
{
"mcpServers": {
"creatorscore": {
"command": "npx",
"args": ["-y", "creatorscore-mcp"],
"env": {
"CREATORSCORE_API_KEY": "sk_live_..."
}
}
}
}Restart Claude Desktop. The tools appear under the 🔌 icon.
claude mcp add creatorscore \
-e CREATORSCORE_API_KEY=sk_live_... \
-- npx -y creatorscore-mcpAny MCP-compatible client works the same way: register a stdio server that runs node /path/to/mcp-server/dist/index.js with the CREATORSCORE_API_KEY environment variable set. Consult your client's MCP / custom-connector settings for where to add it.
Once connected, just ask in natural language:
You: Vet @savzuk's footprint on tiktok for a campaign.
AI → (calls score_creator_footprint) Found 3 accounts — tiktok, youtube,
and instagram @savannah_zuk (cross-handle, via her link-in-bio).
Unified CreatorScore: 84/100. Worst risk: [medium] Profanity (tiktok).
You: What's a fair rate for one 30s beauty TikTok, ~200k median views?
AI → (calls calculate_fair_rate) Fair rate: $2,100 · range $1,790–$2,420.
Free — no credits used.score_creator_fresh can spend scrape credits, and it requires explicit confirmation — it won't fire silently.calculate_fair_rate computes a number.