QueryLines Run a live query

Docs

Use from your assistant

QueryLines runs an MCP server at https://api.querylines.com/mcp (streamable HTTP) exposing the same data as seven typed tools. Three are free in any client with zero setup; four are paid per call and need an x402-capable client with a funded wallet.

Claude Code

Add the server with one command:

terminal 
$ claude mcp add --transport http querylines https://api.querylines.com/mcp

Then ask it something like "what's Florida E&S homeowners premium doing this year?" and it can call get_premium_sample or list_lines immediately, no further setup.

claude.ai

Add QueryLines as a custom connector:

  1. Open Settings.
  2. Go to Customize → Connectors.
  3. Click Add custom connector.
  4. Enter a name (e.g. QueryLines) and the URL https://api.querylines.com/mcp.

On a Team or Enterprise plan, connectors are added org-wide from Organization settings → Connectors by an owner, then enabled per member.

ChatGPT

ChatGPT connects to custom remote MCP servers through Developer mode, available on paid plans (Plus, Pro, Business, Enterprise, Edu):

  1. In Settings, enable Developer mode (currently under Security and login; OpenAI moves this toggle around, so search settings for "developer" if it isn't there). On Business, Team, or Enterprise plans a workspace admin has to allow it first.
  2. Add a new connector with a name and the URL https://api.querylines.com/mcp. No authentication fields are needed.
  3. Enable it in a chat and ask a market question; the three free tools work immediately.

The regular Connectors surface (without Developer mode) only accepts servers built for deep research (search/fetch tools), so it can't use QueryLines' tool set; Developer mode is the supported path.

Microsoft 365 Copilot

Microsoft 365 Copilot reaches MCP servers through an agent built in Copilot Studio. A maker or admin sets it up once, then publishes the agent so it shows up in Copilot chat for the whole team:

  1. In Copilot Studio, open the agent's Tools page and choose Add a tool → New tool → Model Context Protocol.
  2. Enter a server name, the URL https://api.querylines.com/mcp, and a short description; the orchestrator uses the description to decide when to call it, so something like "U.S. surplus lines market statistics by state, month, and coverage line" works well.
  3. Authentication: None. Create the connection and add the tool to the agent.
  4. Publish the agent to Microsoft 365 Copilot and roll it out to your users.

Two footnotes for IT: Copilot Studio talks to MCP through Power Platform connectors, so tenant data-loss-prevention policies apply to it like any other connector. And the Microsoft 365 admin center's "federated connector" route requires Entra SSO or OAuth on the server, which QueryLines doesn't use (no accounts), so Copilot Studio is the path.

GitHub Copilot

In VS Code, create .vscode/mcp.json in your workspace (or use the MCP: Open User Configuration command for a global config). Note the key is servers here, not mcpServers:

.vscode/mcp.json 
{
  "servers": {
    "querylines": {
      "type": "http",
      "url": "https://api.querylines.com/mcp"
    }
  }
}

The tools show up in Copilot Chat's Agent mode (not Ask or Edit). From the Copilot CLI it's one command:

terminal 
$ copilot mcp add --transport http querylines https://api.querylines.com/mcp

The cloud Copilot coding agent on github.com also takes MCP servers (repo Settings → Copilot → MCP servers, with a tools allowlist), but it doesn't support servers that require OAuth and only exposes tools, so treat it as experimental for now.

Cursor & generic MCP clients

Any client that reads a standard mcpServers config (Cursor and most others) takes this block. Set "type": "http", not stdio:

mcp.json 
{
  "mcpServers": {
    "querylines": {
      "type": "http",
      "url": "https://api.querylines.com/mcp"
    }
  }
}

More clients

Any MCP client that speaks streamable HTTP works with the bare URL, no auth. The config key just varies by client:

Client Where Detail
Windsurf ~/.codeium/windsurf/mcp_config.json Uses "serverUrl" instead of "url".
Gemini CLI settings.json → mcpServers Use "httpUrl" for streamable HTTP ("url" means SSE there).
Goose goose configure Add Extension → Remote (Streamable HTTP), paste the URL.

Tools and prices

Tool Price What it does
list_lines free Coverage-line taxonomy (source codes, names, categories) for a state.
get_coverage free Which states, date ranges, and granularities have data available.
get_premium_sample free Statewide, all-lines, monthly premium and policy-count series, latest 13 months.
get_premium $0.10 Full premium/policy-count time series, optionally by coverage line, date range, granularity.
get_summary $0.10 Year-over-year premium and policy-count summary for a state and year.
get_top_insurers $0.10 Top insurers by premium for a state and period.
compare_states $0.50 Premium/policy-count series across multiple states side by side.
Matches https://api.querylines.com/openapi.json and the live tool list at https://api.querylines.com/mcp.

Free tools work in any MCP client, immediately. Plain Claude Desktop, claude.ai, or Cursor users without a wallet configured can call list_lines, get_coverage, and get_premium_sample right after adding the server, no payment setup at all.

Paid tools return an x402 payment challenge instead of data until the calling client can settle it. That means the client itself needs x402 support and a funded wallet, the same USDC-on-Base wallet described in Pay per query: Step 0. An x402-aware agent framework or an MCP proxy like x402-proxy can add that to a client that doesn't support it natively.

Where it's listed

QueryLines is registered in the places agents look for MCP servers: