Getting started

• 1 Authentication — API keys and headers
• 2 Plans and limits — speed, daily volume, portfolio size, history, and how 429s behave
• 3 Response format — payload structure, variants, provenance
• 4 Data quality — freshness, gaps, revisions, what is public about each metric
• 5 Backtesting — point-in-time history without look-ahead

60-second start

# 1. list what your plan can access
curl -H "X-API-Key: YOUR_API_KEY" "https://api.madjik.io/v1/catalog"

# 2. latest value of one metric
curl -H "X-API-Key: YOUR_API_KEY" "https://api.madjik.io/v1/metrics/M10004"

# 3. its history with provenance
curl -H "X-API-Key: YOUR_API_KEY" \
  "https://api.madjik.io/v1/metrics/M10004/timeseries?interval=1h"

Prefer a client library, or an AI assistant?

# Python SDK (see /sdks/ for the install command)
from madjik import Madjik
Madjik().metric("M10004")

// TypeScript SDK
await new Madjik().metric("M10004");

The SDKs handle retries, rate limits and history paging; the MCP server lets an AI assistant call the same data as a tool.