Why I Built a CLI for Massive Financial Data

I've been using the Massive financial data API for a while now. It's a solid API — real-time and historical market data across stocks, options, crypto, forex, futures, and indices. But every time I wanted to pull data, I was either writing throwaway scripts or wiring up an MCP server so my AI agent could fetch it for me.

Both approaches had problems. Scripts are disposable. MCP servers flood your agent's context window with massive JSON payloads, eating up tokens and degrading response quality. And when something breaks, debugging is opaque — you can't see what the agent actually sent or received.

So I built a CLI instead. I call it Massive CLI.

---

Why a CLI Instead of an MCP Server?

This is the question I keep getting, so let me address it head-on. There are four reasons I chose a CLI over an MCP server:

---

What It Covers

Massive CLI wraps the entire Massive API into a single binary. That means access to six major asset classes, real-time WebSocket streaming, bulk historical data downloads, and partner data sources — all from your terminal.

On top of that, there's partner data from Benzinga (news, ratings, earnings), economic data (inflation, labor market, treasury yields), ETF analytics, and Canadian market data via TMX.

---

Getting Started

Installation is straightforward. If you're on macOS or Linux, Homebrew is the easiest path:

Then configure your API key:

That's it. You can also set MASSIVE_API_KEY as an environment variable or drop it in a .env file — whatever fits your workflow. Pre-built binaries for macOS, Linux, and Windows are available on the releases page if you don't use Homebrew.

---

How It Works in Practice

The command structure is intuitive. You pick an asset class, pick a command, and pass in your parameters. Here are some examples I use daily:

Pull OHLC Bars for a Stock

This gives you a clean table of daily bars. Want weekly bars instead? Add --timespan week. Want JSON? Add -o json.

Check Today's Market Movers

Get Company Fundamentals

Stream Real-Time Trades

The WebSocket streaming is particularly useful. You can subscribe to trades, quotes, or minute/second aggregates across any asset class. By default you get 15-minute delayed data — add --realtime if your subscription supports it.

Download Bulk Historical Data

This pulls gzipped CSV files from Massive's S3-compatible storage — perfect for backtesting or building research datasets.

---

The AI Agent Use Case

This is the use case I'm most excited about. I use Claude Code daily, and having a CLI that Claude can shell out to is far more practical than an MCP server for financial data.

Here's why: when Claude needs stock data, it runs a command and gets back a focused result. It doesn't need to parse a massive JSON blob or manage an open connection. It just runs a command like any other tool:

The --output json flag is the key. Table output is great for humans; JSON output is great for agents and scripts. Same command, same data, different format. No adapter layer needed.

---

Under the Hood

The tool is written in Go and built on Cobra — the same CLI framework used by Kubernetes, Hugo, and GitHub's CLI. The architecture is clean and predictable:

• REST client — A straightforward HTTP client that authenticates with your API key and returns typed Go structs. Every command follows the same pattern: build params, call the API, render output as a table or JSON.
• WebSocket client — Uses gorilla/websocket for streaming. Connects, subscribes to channels (trades, quotes, aggregates), and delivers messages until you hit Ctrl+C.
• S3 client — Uses the AWS SDK to connect to Massive's S3-compatible endpoint for flat file downloads. List files by year/month, download gzipped CSVs by date.
• Configuration — API keys stored at ~/.config/massive/config.json with 0600 permissions. Environment variables take priority over the config file. Supports .env files.

The whole project compiles to a single binary with no external runtime dependencies. CI/CD automatically cross-compiles for macOS (Intel and Apple Silicon), Linux, and Windows on every push to main.

---

The Full Command Tree

To give you a sense of how comprehensive the coverage is, here's the full command tree. Every endpoint in the Massive API is accessible from the terminal:

---

Give It a Try

The project is open source and available on GitHub: github.com/cloudmanic/massive. If you have a Massive API key, you can be pulling data in under a minute.

If you're using AI coding assistants like Claude Code and need financial data, I think you'll find this approach — a simple CLI with table and JSON output — far more practical than an MCP server. Your agent gets exactly the data it needs without blowing up its context window, and you can debug any issue by running the same command yourself.

Star the repo, try it out, and open an issue if you run into anything. I'm actively developing this and shipping updates regularly.