Quickstart

This guide uses Direct Execute mode with Meridian pinned by tool ID. Execute returns the typed market-data response directly and charges the method’s published per-call price. Query mode remains available when Context should interpret a natural-language question.

1. Create a Context API key

Sign in at ctxprotocol.com.
Configure the wallet spending cap required by Context.
Fund the Context wallet for paid tool calls.
Create an API key under Settings → API Keys.

Store the key in an environment variable. Never embed it in browser code or commit it to a repository.

export CONTEXT_API_KEY="sk_live_YOUR_KEY"

2. Install an SDK

npm install @ctxprotocol/sdk

pip install ctxprotocol

3. Execute Meridian

import { ContextClient } from "@ctxprotocol/sdk";

const MERIDIAN_TOOL_ID = "e5e62fa5-28e1-42f4-8b84-f4866234df43";
const client = new ContextClient({
  apiKey: process.env.CONTEXT_API_KEY!,
});

const meridian = await client.discovery.get(MERIDIAN_TOOL_ID);
const futures = meridian.mcpTools?.find((method) => method.name === "futures");
if (!futures?.executeEligible || !futures.executePriceUsd) {
  throw new Error("Meridian futures is not currently Execute-eligible");
}

const result = await client.tools.execute({
  toolId: MERIDIAN_TOOL_ID,
  toolName: "futures",
  args: {
    exchanges: ["binance", "bybit", "okx", "hyperliquid"],
    products: ["BTCUSDT"],
    columns: ["close_price", "funding_rate", "dollar_open_interest_close"],
    begin: "2026-08-06T23:00:00Z",
    end: "2026-08-07T00:00:00Z",
    resolution: "1h",
    include: {
      coverage: false,
      provenance: false,
      column_metadata: false
    }
  },
  maxSpendUsd: "0.005",
  closeSession: true,
});

console.log(result.result.rows);

import asyncio
import os
from uuid import uuid4

from ctxprotocol import ContextClient

MERIDIAN_TOOL_ID = "e5e62fa5-28e1-42f4-8b84-f4866234df43"

async def main() -> None:
    async with ContextClient(api_key=os.environ["CONTEXT_API_KEY"]) as client:
        meridian = await client.discovery.get(MERIDIAN_TOOL_ID)
        futures = next(
            (method for method in (meridian.mcp_tools or []) if method.name == "futures"),
            None,
        )
        if not futures or not futures.execute_eligible or not futures.execute_price_usd:
            raise RuntimeError("Meridian futures is not currently Execute-eligible")

        result = await client.tools.execute(
            tool_id=MERIDIAN_TOOL_ID,
            tool_name="futures",
            args={
                "exchanges": ["binance", "bybit", "okx", "hyperliquid"],
                "products": ["BTCUSDT"],
                "columns": [
                    "close_price",
                    "funding_rate",
                    "dollar_open_interest_close",
                ],
                "begin": "2026-08-06T23:00:00Z",
                "end": "2026-08-07T00:00:00Z",
                "resolution": "1h",
                "include": {
                    "coverage": False,
                    "provenance": False,
                    "column_metadata": False,
                },
            },
            idempotency_key=str(uuid4()),
            max_spend_usd="0.005",
            close_session=True,
        )
        print(result.result["rows"])

asyncio.run(main())

4. Inspect the response

result.result["rows"] is a DataFrame-ready list of normalized row objects. Each Execute response also reports method and session billing metadata. Meridian rows use canonical exchange, product, and time identities. Optional response sections include:

columns: names, types, units, and data families.
coverage: what exists for each exchange/product/range.
provenance: the source and collection lineage.
warnings: explicit limitations or source-limited values.

Do not infer missing values as zero. For a quality audit, repeat the narrow query with coverage=true and classify returned nulls against the coverage declarations.

Query mode

Use pinned Query mode when Context should select methods, verify evidence, and synthesize a natural-language answer:

result = await client.query.run(
    query="Compare BTC funding and dollar open interest across futures venues.",
    tools=[MERIDIAN_TOOL_ID],
    response_shape="evidence_only",
    include_data=True,
)

Overview

Getting Started

Context Integration

Tool Reference

Data Catalog

Custom Bars

Operations

1. Create a Context API key

2. Install an SDK

3. Execute Meridian

4. Inspect the response

Query mode

Next steps

TypeScript integration

Python integration

Official MCP endpoint

Futures reference

Pricing and limits

​1. Create a Context API key

​2. Install an SDK

​3. Execute Meridian

​4. Inspect the response

​Query mode

​Next steps

TypeScript integration

Python integration

Official MCP endpoint

Futures reference

Pricing and limits

1. Create a Context API key

2. Install an SDK

3. Execute Meridian

4. Inspect the response

Query mode

Next steps