MCP · Model Context Protocol
Creator intelligence, right inside your AI assistant
Connect Zaver Radar to Claude, Cursor, ChatGPT and any MCP client, and ask for creators the way you’d ask a colleague. 26 tools for discovery, insights, competitive intelligence and social listening — answered in conversation, priced in the same credits as the app.
Read the MCP docs →Get an API key
Connecting is free · tool calls use your workspace credits · needs the Developer API add-on
What it is
The Model Context Protocol lets an AI assistant call external tools. Radar’s MCP server turns its creator brain into tools your assistant can use directly: describe the creators you need and it runs a discovery search; paste a name and it returns a full insight report; ask “how does Glossier compare to Rare Beauty?” and it runs the competitor comparison — all without leaving the chat.
Endpoint: https://radar.zaver.one/mcp · Auth: OAuth 2.1 (PKCE) or Bearer pk_live_… keys from Settings → Developer API.
Connect it in a minute
Claude (web, desktop, mobile)
Settings → Connectors → Add custom connector, name it “Zaver Radar”, paste the MCP URL https://radar.zaver.one/mcp, then sign in with OAuth.
Claude Code
claude mcp add --transport http zaver-radar https://radar.zaver.one/mcpCursor
// ~/.cursor/mcp.json
{
"mcpServers": {
"zaver-radar": {
"url": "https://radar.zaver.one/mcp"
}
}
}ChatGPT, VS Code, Codex CLI, Windsurf
ChatGPT needs Developer mode (paid plans): Settings → Apps & Connectors → Advanced, create a custom connector with OAuth. VS Code, Codex CLI and Windsurf follow the same HTTP-transport pattern — full config for each is in the MCP docs.
26 tools, grouped by job
A representative set of what your assistant can call. Read-only access is available; full access is required for searches and enrichment.
discover_creatorssearch_creator_databasefind_creator_by_name
get_creator_insightsget_creator_profileget_creator_posts
decode_brand_strategyget_competitor_comparisontrack_competitor
get_listening_summarysearch_listening_mentionsget_ugc_creators
enrich_creator_contactsrun_data_tool
get_accountget_job_status
Full tool reference, parameters and examples are in the MCP docs.
Pricing & access
- Connecting is free —
get_accountshows your balance and per-tool costs - Tool calls spend the same credits as the app (discovery, insights, data lookups)
- Requires the Developer API add-on on your workspace plan
- Every call is logged in Settings → Audit log; long jobs return an ID for
get_job_status
New to Radar? See plans · Developer API overview
Frequently asked questions
What is the Zaver Radar MCP server?
It’s a Model Context Protocol server that exposes Radar’s creator search, vetting and social-intelligence capabilities as 26 tools your AI assistant can call — so you can find, vet and price creators, decode competitors and monitor mentions in plain conversation.
What is the MCP endpoint?
https://radar.zaver.one/mcp — add it as a custom connector (Claude), a CLI command (Claude Code), or an entry in your client’s mcp.json (Cursor and others).
How does authentication work?
Sign-in connections use OAuth 2.1 with PKCE; programmatic use takes an Authorization: Bearer pk_live_… key created under Settings → Developer API. Tokens expire after 1 hour with a 30-day refresh.
What does it cost?
Connecting is free. Tool calls spend the same credits as the app — discovery runs, insight reports and data lookups at your plan’s rates. Call get_account to see your balance and per-tool costs. It requires the Developer API add-on on your workspace plan.
Is usage tracked?
Yes — every MCP tool call is logged in Settings → Audit log, and long-running tasks return a job ID you check with get_job_status.
Which assistants can connect?
Claude (web, desktop, mobile), Claude Code, Cursor, ChatGPT (Developer mode), VS Code, Codex CLI and Windsurf — any client that speaks MCP over HTTP.