DEV Community: a10102010

Build a Trading Agent in 10 Lines of Python

a10102010 — Fri, 16 Oct 2026 18:43:31 +0000

A ready-to-run AI trading agent that scans 100+ assets, checks geopolitical risk, and makes BUY/SELL decisions — powered by real market data.

Last month, my AI agent told me to buy Sugar futures. Three weeks later: +16.5%.

I didn't write a single line of trading logic. The agent did everything — checked geopolitical risk, analyzed weather anomalies i Brazil, cross-referenced USDA crop data, and decided to go long. I just ran the script.

Here's how to build the same agent in 10 lines of Python.

The Agent

from north7 import North7Client

client = North7Client() # free: no key needed for risk/regime

risk = client.risk()

regime = client.regime()

print(f"Risk: {risk.risk_index}/100 ({risk.label})")

print(f"Regime: {regime.regime}")

signals = client.signals() # needs API key

best = max(signals, key=lambda s: s.risk_score or 0)

print(f"Top signal: {best.ticker} {best.action} (score {best.risk_score})")

That's it. Install with pip install north7, run it, get a trading decision.

What it actually does

Behind those 10 lines, the API scans 80+ assets across stocks, commodities, crypto, and forex. It pulls from:

70+ data sources in 30+ languages
NOAA weather data (drought, flooding, harvest impact)
USDA crop reports and EIA energy stats
Geopolitical intelligence (sanctions, trade wars, conflicts)
Technical momentum and sector rotation

The Risk Index (0-100) tells you how dangerous the market is right now. The Regime (BULL/BEAR/SIDEWAYS/CRISIS) tells you the trend.

Signals tell you what to trade.

The full agent

The complete trading agent adds AI analysis and decision logic:

from north7 import North7Client

client = North7Client()

# 1. Market context

risk = client.risk()

regime = client.regime()

# 2. Get signals + pick the best

signals = client.signals()

best = max(signals, key=lambda s: s.risk_score or 0)

# 3. AI deep-dive on the top pick

analysis = client.analysis(best.ticker)

# 4. Decision

if best.risk_score >= 70 and best.action == "BUY":

print(f"BUY {best.ticker} — {analysis.summary[:200]}")

elif best.risk_score >= 70 and best.action == "SELL":

print(f"SELL {best.ticker}")

else:

print("HOLD — no strong signal")

Run it as a cron job

# Every morning at 08:00 UTC

0 8 * * * python agent.py >> trades.log 2>&1

Or as a webhook agent that sends Telegram alerts:

from north7 import North7Client

import requests, time

client = North7Client()

while True:

signals = client.signals()

for s in signals:

if s.risk_score >= 75:

requests.post(f"https://api.telegram.org/bot{TOKEN}/sendMessage",

data={"chat_id": CHAT, "text": f"ALERT: {s.action} {s.ticker} (score {s.risk_score})"})

time.sleep(300) # check every 5 min

Free endpoints (no API key)

You can start without signing up:

# Risk Index — how dangerous is the market right now?

curl https://north7.ai/v1/risk

# Market Regime — BULL, BEAR, SIDEWAYS, or CRISIS?

curl https://north7.ai/v1/regime

For signals and analysis, get a free API key — 1,000 credits, no credit card.

MCP for AI agents

If you're building with Claude, GPT, or any LLM that supports Model Context Protocol:

{

"mcpServers": {

"north7": {

"command": "npx",

"args": ["-y", "@anthropic/mcp-remote", "https://north7.ai/v1/mcp"]

}

}

}

Your AI agent can now call get_risk_index(), get_signals(), get_analysis() etc. as tools.

Why this works

Most financial APIs give you raw data — price feeds, candlesticks, order books. You still need to build the logic that turns data

into decisions.

NORTH7 gives you decision-ready signals. The AI has already analyzed the data, cross-referenced sources, and assigned confidence

scores. Your agent just needs to act on them.

The Sugar trade? The agent spotted a +174% rainfall anomaly in Sao Paulo (threatening Brazil's harvest) combined with -99%

precipitation in India. It cross-referenced historical patterns and found that similar setups led to +8-16% moves within 3 weeks. The
signal came out at $17.40. Sugar hit $20.27.

Get started

pip install north7

GitHub: ready-to-run examples
API Documentation
Python SDK on PyPI

tags: python, ai, trading, api

I built a financial intelligence API for AI agents — here's what happened

a10102010 — Sat, 15 Aug 2026 20:42:57 +0000

Most financial APIs give you raw data. Price feeds. OHLCV candles. Maybe some moving averages.

But if you're building an AI agent that needs to make decisions — not just display charts — raw data isn't enough. Your agent needs to know: Is the market dangerous right now? Should I buy or sell? What's happening geopolitically that could move prices tomorrow?

That's what I built. Here's the story.

The problem

I run a trading bot. It connects to Interactive Brokers and trades stocks, commodities, and forex autonomously. But the bot was only as good as the signals I fed it.

I had 6 different signal sources — news analysis, alpha screening, stock radar, commodity scoring, an intraday scanner, and a geopolitical risk system analyzing OSINT in 15+ languages. They all generated good data. But none of them talked to each other, and none of them were accessible to external agents or tools.

So I built an API layer on top of everything.

What the API does

Three endpoints are completely free — no API key needed:

# Global risk index (0-100)
curl https://north7.ai/v1/risk

# Market regime (BULL / BEAR / SIDEWAYS / CRISIS)
curl https://north7.ai/v1/regime

# API health + data freshness
curl https://north7.ai/v1/health

Try them right now. You'll get a response like this:

{
  "ok": true,
  "data": {
    "value": 88,
    "status": "HIGH",
    "crises": [
      "Strait of Hormuz blockade unresolved",
      "Russia-Ukraine drone strikes on energy infrastructure",
      "Colombia earthquake aftermath"
    ],
    "regions": [
      {"name": "Middle East", "score": 91},
      {"name": "Eastern Europe", "score": 87}
    ]
  }
}

That's not raw data. That's decision intelligence. An AI agent reads this and knows: risk is elevated, be cautious, reduce exposure.

Get a free API key in 10 seconds

curl -X POST "https://north7.ai/v1/checkout/free?email=you@example.com&name=MyAgent"

You get 1,000 credits instantly. No credit card. No approval process. No human in the loop.

Then you can access all 14 endpoints:

Endpoint	Credits	What it does
`/v1/signals/latest`	1	66 trading signals from 5 AI sources
`/v1/signals/alpha`	2	Highest-conviction daily picks
`/v1/prices`	1	Real-time prices for 80+ assets
`/v1/analysis/{ticker}`	5	Deep AI analysis with fundamentals
`/v1/intelligence/briefing`	10	Daily geopolitical briefing
`/v1/supply-chain/risk`	2	Supply chain risk for 37 commodities
`/v1/supply-chain/alerts`	2	High-risk commodity alerts
`/v1/sector-radar`	2	Sector rotation analysis
`/v1/commodities`	2	Commodity scoring (37 materials)
`/v1/intraday`	2	Intraday momentum scanner

Full docs: north7.ai/v1/docs

MCP integration (for Claude users)

If you use Claude Desktop or any MCP-compatible client, you can add NORTH7 as a tool source:

{
  "mcpServers": {
    "north7": {
      "url": "https://north7.ai/mcp",
      "transport": "streamable-http"
    }
  }
}

Claude will then have 9 financial intelligence tools available:

get_risk_index — Is the market dangerous?
get_market_regime — Bull or bear?
get_trading_signals — What should I trade?
get_prices — Current prices
get_stock_analysis — Deep dive on a ticker
get_intelligence_briefing — What's happening in the world?
get_commodity_signals — Raw material risk scores
get_sector_radar — Which sectors are leading?
get_model_portfolio — What the AI portfolio holds

No API key needed for discovery. Your Claude agent can explore the tools, see what's available, and then get a free key to start using them.

Supply chain risk — the unexpected use case

One endpoint I didn't plan for became the most interesting: supply chain risk.

curl "https://north7.ai/v1/supply-chain/alerts?threshold=80" \
  -H "X-API-Key: your_key"

{
  "alerts": [
    {
      "commodity": "Brent Crude",
      "risk_score": 91,
      "action": "HEDGE",
      "recommendation": "Lock in forward contracts for 60-75% of Brent-linked supply..."
    },
    {
      "commodity": "Sugar",
      "risk_score": 85,
      "action": "BUY",
      "recommendation": "Accelerate procurement..."
    }
  ]
}

37 commodities. Real-time risk scoring across 5 dimensions: geopolitics, weather, logistics, currency, and price trends. Each with a concrete recommendation: BUY, HEDGE, WATCH, or AVOID.

This isn't for traders. This is for procurement teams who need to know if their wheat supply from Ukraine is at risk, or if coffee prices are about to spike because of drought in Brazil.

What I learned building this

1. AI agents don't search for APIs. Not yet. Humans configure agents with tools. So distribution is a human problem, not a technical one.

2. Free endpoints are the hook. Risk and regime are free. Agents try them, see value, get a key. Without free endpoints, nobody tries.

3. Error messages are documentation. When an agent hits a 401, the response tells them exactly where to get a key and what free endpoints exist. Every error is a conversion opportunity.

4. The next field changes everything. Every response includes a next suggestion — the logical next API call. This guides agents through a workflow instead of leaving them guessing.

"meta": {
  "next": {
    "endpoint": "/v1/signals/latest",
    "reason": "Get trading signals aligned with the current regime"
  }
}

5. MCP is early but important. Most traffic comes from humans reading docs, not agents auto-discovering via MCP. But the infrastructure needs to be there for when autonomous agents arrive.

The numbers

24 API endpoints
9 MCP tools
80+ tracked assets
37 commodities with supply chain risk
1,071 verified geopolitical predictions (52.2% calibrated accuracy)
Response time: ~500ms
Free tier: 1,000 credits

Try it

# No key needed
curl https://north7.ai/v1/risk
curl https://north7.ai/v1/regime

# Get free key
curl -X POST "https://north7.ai/v1/checkout/free?email=you@example.com"

# Use it
curl https://north7.ai/v1/signals/latest -H "X-API-Key: your_key"

API docs: north7.ai/v1/docs
MCP server: north7.ai/mcp
Developer page: north7.ai/developers
GitHub: github.com/a10102010/north7-agent-api

Built in Austria. Open for feedback.