Give Your AI Agent Real Tools — 40 APIs Behind One Key

#ai #api #python #tutorial

Your AI agent can write poetry and explain quantum physics. But can it take a screenshot? Check a crypto price? Run a Python script? Look up a DNS record?

Probably not. Because connecting an LLM to real-world tools is annoying:

Find 10 different APIs
Sign up for 10 different accounts
Manage 10 different API keys
Handle 10 different rate limits and auth schemes

What if you could give your agent 40 real tools with one API call?

One API Key, 40+ Tools

curl -X POST https://agent-gateway-kappa.vercel.app/api/keys/create

That returns an API key with 200 free credits. No email, no signup form, no credit card. Now your agent can:

import requests

API_KEY = "gw_your_key_here"
BASE = "https://agent-gateway-kappa.vercel.app/v1"
HEADERS = {"Authorization": f"Bearer {API_KEY}"}

# Take a screenshot of any website
r = requests.post(f"{BASE}/agent-screenshot/api/screenshot",
    headers=HEADERS,
    json={"url": "https://news.ycombinator.com", "viewport": "desktop"})
screenshot_url = r.json()["url"]

# Look up an IP address
r = requests.get(f"{BASE}/agent-geo/api/geo/8.8.8.8", headers=HEADERS)
location = r.json()  # country, city, timezone, coordinates

# Check crypto prices
r = requests.get(f"{BASE}/crypto-feeds/api/price/BTC", headers=HEADERS)
btc_price = r.json()["price"]

# Execute Python code in a sandbox
r = requests.post(f"{BASE}/agent-coderunner/api/execute",
    headers=HEADERS,
    json={"language": "python", "code": "print(2**256)"})
output = r.json()["output"]

# Scrape a webpage to markdown
r = requests.post(f"{BASE}/agent-scraper/api/scrape",
    headers=HEADERS,
    json={"url": "https://example.com", "format": "markdown"})
content = r.json()["content"]

Five different capabilities. One API key. One base URL. Same auth header everywhere.

The Full Tool List

Here's what your agent gets access to:

Data & Research

IP Geolocation — country, city, timezone, ISP for any IP
Web Scraper — extract content as markdown, HTML, or plain text
Web Search — DuckDuckGo-powered search results
Crypto Price Feeds — 500+ tickers from Binance, real-time
DNS Lookup — resolve records, WHOIS, domain availability
On-Chain Analytics — trending tokens, gas prices, wallet data

Actions

Screenshot — render any URL to PNG (desktop/mobile/tablet)
Code Runner — execute Python, JavaScript, TypeScript, Bash
PDF Generator — HTML/Markdown to PDF
Email Sender — send emails with templates
URL Shortener — create tracked short links
QR Code Generator — generate QR codes from text

Infrastructure

File Storage — upload/download with expiring links
Task Queue — submit and process background jobs
Scheduler — cron jobs and one-time tasks
Event Bus — pub/sub messaging
Webhook Relay — receive, inspect, and forward webhooks
Key-Value Store — persist data between agent runs
Vector Store — semantic search over documents
Log Drain — centralized logging

Blockchain

Multi-Chain Wallet — 9 blockchains, balance checks, swaps
Contract Deployer — deploy Solidity contracts
Token Launchpad — create ERC-20 tokens
DeFi Trading — 275+ perpetual futures markets

Building a Simple AI Agent

Here's a minimal agent that uses tools based on natural language:

import requests
import json

API_KEY = "gw_your_key_here"
BASE = "https://agent-gateway-kappa.vercel.app/v1"
H = {"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"}

def tool_screenshot(url):
    r = requests.post(f"{BASE}/agent-screenshot/api/screenshot",
        headers=H, json={"url": url})
    return r.json().get("url", "Failed")

def tool_geoip(ip):
    r = requests.get(f"{BASE}/agent-geo/api/geo/{ip}", headers=H)
    d = r.json()
    return f"{d.get('city','?')}, {d.get('country','?')} ({d.get('timezone','?')})"

def tool_crypto_price(symbol):
    r = requests.get(f"{BASE}/crypto-feeds/api/price/{symbol}", headers=H)
    return f"${r.json().get('price', '?')}"

def tool_dns(domain):
    r = requests.get(f"{BASE}/agent-dns/api/resolve/{domain}", headers=H)
    return json.dumps(r.json().get("answers", []), indent=2)

def tool_scrape(url):
    r = requests.post(f"{BASE}/agent-scraper/api/scrape",
        headers=H, json={"url": url, "format": "markdown"})
    return r.json().get("content", "Failed")[:2000]

# Map tool names to functions
TOOLS = {
    "screenshot": tool_screenshot,
    "geoip": tool_geoip,
    "crypto": tool_crypto_price,
    "dns": tool_dns,
    "scrape": tool_scrape,
}

# Simple dispatcher
def run_tool(name, arg):
    if name in TOOLS:
        return TOOLS[name](arg)
    return f"Unknown tool: {name}"

# Example: an agent loop
tasks = [
    ("screenshot", "https://news.ycombinator.com"),
    ("geoip", "8.8.8.8"),
    ("crypto", "ETH"),
    ("dns", "github.com"),
]

for tool_name, arg in tasks:
    result = run_tool(tool_name, arg)
    print(f"[{tool_name}] {arg} → {result}")

What It Costs

200 free credits on signup (1 credit = 1 API call)
No expiration pressure — use them whenever
Need more? $1 USDC = 500 additional credits
Pay with USDC on Base or XMR — no credit card needed

Check your balance anytime:

curl -H "Authorization: Bearer YOUR_KEY" \
  https://agent-gateway-kappa.vercel.app/api/keys/balance

Try It Now

Get your key and start building:

# 1. Get your API key
curl -s -X POST https://agent-gateway-kappa.vercel.app/api/keys/create | python3 -m json.tool

# 2. Take a screenshot (replace YOUR_KEY)
curl -s -X POST https://agent-gateway-kappa.vercel.app/v1/agent-screenshot/api/screenshot \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com"}' | python3 -m json.tool

# 3. Check your remaining credits
curl -s -H "Authorization: Bearer YOUR_KEY" \
  https://agent-gateway-kappa.vercel.app/api/keys/balance | python3 -m json.tool