MCP server overview & quickstart

How it will work

• One remote endpoint: https://review.expert/mcp, using the MCP Streamable HTTP transport. Clients send JSON-RPC over POST; the server may stream responses with server-sent events. Nothing to install locally.
• Two ways to authenticate: OAuth 2.1 with Dynamic Client Registration, so clients like Claude and ChatGPT can register and sign you in, or a team API key (rx_live_…) sent as a Bearer token. See authentication.
• Your team's data only: every request is scoped to one Review Expert team and the apps it has connected.
• Read-only by default: tools read reviews and stats. A reply-draft tool is planned for a later phase and will never post to a store.

Planned tools

• list_apps — List the App Store and Google Play apps connected to your team.
• get_app_reviews — Fetch reviews for one app, filtered by date, rating, territory and reply status.
• search_reviews — Full-text search across reviews for all of your team's apps.
• review_stats — Counts, average rating, rating distribution and response rate for a period.
• sentiment_trends — Rating-based sentiment buckets (positive/neutral/negative) over time.
• draft_reply — Return a suggested reply to a review. Never posts it (later phase).

Full input and output schemas are in the tool reference.

Quickstart (planned configuration)

Each snippet shows how a client would connect once early access opens. If you use an API key, replace rx_live_… with a key from Admin → Settings → API keys (planned).

Claude Code

# OAuth (opens a browser to sign in and approve scopes)
claude mcp add --transport http reviewexpert https://review.expert/mcp

# Or with a team API key
claude mcp add --transport http reviewexpert https://review.expert/mcp \
  --header "Authorization: Bearer rx_live_…"

Claude Desktop and claude.ai

Remote MCP servers are added as a custom connector: Settings → Connectors → Add custom connector, then enter https://review.expert/mcp as the URL. Claude will run the OAuth sign-in.

claude_desktop_config.json launches local processes, so to configure the server there you can use the mcp-remote bridge:

{
  "mcpServers": {
    "reviewexpert": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://review.expert/mcp"]
    }
  }
}

ChatGPT

1. Open Settings → Apps & Connectors → Advanced settings and turn on Developer mode (availability depends on your ChatGPT plan).
2. Create a connector and enter https://review.expert/mcp as the MCP server URL with OAuth authentication.
3. Sign in to Review Expert and approve the requested scopes.

Cursor

{
  "mcpServers": {
    "reviewexpert": {
      "url": "https://review.expert/mcp",
      "headers": {
        "Authorization": "Bearer rx_live_…"
      }
    }
  }
}

VS Code

{
  "servers": {
    "reviewexpert": {
      "type": "http",
      "url": "https://review.expert/mcp"
    }
  }
}

Any Streamable HTTP client

curl -X POST https://review.expert/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "Authorization: Bearer rx_live_…" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "initialize",
    "params": {
      "protocolVersion": "2025-11-25",
      "capabilities": {},
      "clientInfo": { "name": "my-agent", "version": "1.0.0" }
    }
  }'

# The response may include an Mcp-Session-Id header. Send it back on
# later requests, then call tools/list and tools/call.

Treat review text as data

Review titles and bodies are written by the public. Your agent should treat them as untrusted content, not as instructions. The server will label review text accordingly and limit response sizes.

Next steps

• Tool reference — schemas and example calls.
• Authentication — OAuth 2.1, Dynamic Client Registration, scopes and API keys.
• Limits & pricing — proposed plans and rate limits.