← BackRun your report →

Give your AI agent verified market research, not hallucinated numbers

RaportAgent's MCP server puts the same seven-agent, source-checked research engine behind the product directly inside Claude Desktop, Claude Code, Cursor or any other MCP client — so an agent can pull a sourced report instead of guessing from its own training data.

RaportAgent ships an official Model Context Protocol server: @raportagent/mcp. Any MCP client — Claude Desktop, Claude Code, Cursor — can call it to generate a sourced market report, then pull the finished markdown and its full audit trail. It's a thin wrapper over the same REST API the web app uses, so it inherits the same credit model, rate limits and citation guarantees — nothing is invented for the agent that isn't already true for a human buyer.

Why an agent should use this, not just answer from its own knowledge

Tools

ToolWhat it does
generate_reportStart a report (query, optional template: compliance/pitch/saas/ecommerce/realestate/local/battlecard/duediligence/fintech). Returns a report_id; can poll inline. Costs 1 credit (2 for battlecard).
get_report_statusqueued / running / completed / failed
get_reportThe finished report's full markdown and sections
get_report_auditProvenance: models, agents, source counts, SHA-256 of the exact content
get_report_sourcesEvery cited source with link-health counts (working / uncertain / dead)
cancel_reportCancel a queued/in-progress report and refund its credit
list_reportsRecent reports for the authenticated key
get_accountRemaining credits and plan

Setup

  1. Get an API key: create a free account, then My account → API keys.
  2. No install or build step — npx pulls the published package on first run.
{
  "mcpServers": {
    "raportagent": {
      "command": "npx",
      "args": ["-y", "@raportagent/mcp"],
      "env": { "RAPORTAGENT_API_KEY": "ra_live_xxx" }
    }
  }
}

Drop that into claude_desktop_config.json, register it with claude mcp add for Claude Code, or the equivalent MCP settings in Cursor. Full instructions and source: npm / GitHub.

FAQ

What is MCP?

The Model Context Protocol is an open standard for connecting AI assistants to external tools and data. An MCP server exposes a defined set of callable tools — here, market-research generation — that any compatible client (Claude Desktop, Claude Code, Cursor) can invoke directly inside a conversation.

Does it work with Claude Code and Cursor, not just Claude Desktop?

Yes — it's a standard MCP server over stdio, so any MCP-compatible client works. The setup page shows the exact config for Claude Desktop, Claude Code and Cursor.

How much does it cost to use from an agent?

The same credit model as the web app: generating a report costs 1 credit, from €19 per report or less on a plan. Reading status, results, audit trail and account info is free.

Where does my API key get sent?

Only to the base URL you configure (raportagent.com by default). The server is a thin, open-source wrapper over the public REST API — you can read exactly what it does.

Can I self-host or point it at a different base URL?

Yes — set RAPORTAGENT_BASE_URL if you're running against a staging or self-hosted instance of the API.

Wire it into your MCP client

Create a free account to get an API key, then drop the config above into your client — no build step.

Create a free account →
Home · Sample reports by niche · vs ChatGPT & Perplexity · Pricing · FAQ