Composed market and macro data, bearer-token auth, USDC micropayments. No subscription, no KYC, no human in the loop.
USE CASE USDC ON BASE PAY PER CALLIf you're building an autonomous trading bot, you have a payments problem before you have a data problem. Most paid market-data APIs (Messari, CoinAPI, Alpha Vantage paid, Polygon.io) require a human signup, KYC, and a $50-$500 monthly subscription before your bot can fetch a single quote. That's a UX bug for the agent economy. TerminalFeed's premium API tier is built around the opposite assumption: your bot has a wallet, not an email address.
A working trading agent (whether it's an LLM-driven research-and-execute loop, a classic algo bot calling functions, or a multi-agent system orchestrated by something like LangGraph) typically needs all of the following on every decision cycle:
Without TerminalFeed, your bot makes 8-15 separate HTTP calls to 8-15 separate providers, each with its own auth, its own rate-limit pool, its own schema, and its own failure modes. With TerminalFeed, three calls cover the entire decision surface for $0.10 per cycle.
| Endpoint | Cost | Composes |
|---|---|---|
| /api/pro/briefing | $0.02 | BTC + Fear&Greed + earthquakes + HN + Polymarket prediction markets |
| /api/pro/macro | $0.04 | Fed rate + CPI + unemployment + GDP + 10Y treasury + forex (EUR/JPY/GBP/CHF) + commodities (gold, oil, nat gas) + indices (SPY/DIA/QQQ/VIX) |
| /api/pro/crypto-deep | $0.04 | top 50 coins (1h/24h/7d change) + Binance top 20 USDT pairs + mempool.space network stats + Etherscan gas oracle |
Each endpoint absorbs the upstream cost, key management, and rate limits. You bring one bearer token, get one schema, handle one set of failure modes. If an upstream is degraded, we serve stale cache rather than 5xx, so your bot never gets surprised by a `null` quote in the middle of a decision.
Below is a full reference loop. Drop into any agent runtime that exposes requests (which is most of them, including LangChain tool-calling layers, AutoGPT, and custom orchestrators). Replace the strategy function with your actual logic.
import os, time, requests
TF_TOKEN = os.environ["TF_TOKEN"] # tf_live_<64-char-hex>
TF_BASE = "https://terminalfeed.io"
HEADERS = {"Authorization": f"Bearer {TF_TOKEN}"}
def fetch_decision_context():
"""One decision cycle. Costs 4 credits = $0.08 at $1 USDC = 50 credits."""
macro = requests.get(f"{TF_BASE}/api/pro/macro", headers=HEADERS, timeout=10).json()
crypto = requests.get(f"{TF_BASE}/api/pro/crypto-deep?coins=btc,eth", headers=HEADERS, timeout=10).json()
return {"macro": macro, "crypto": crypto}
def strategy(ctx):
btc = ctx["crypto"]["coins_top50"][0]
fed_rate = ctx["macro"]["economic"]["fed_rate"]["value"]
vix = ctx["macro"]["markets"]["vix"]["price"]
# Your trading logic here. This is just an illustrative skeleton.
if btc["change_24h_percent"] < -5 and vix > 25:
return {"action": "buy", "size_usd": 100}
return {"action": "hold"}
def loop():
while True:
ctx = fetch_decision_context()
decision = strategy(ctx)
print(decision, "credits left:", ctx)
time.sleep(300) # 5 min cadence
if __name__ == "__main__":
loop()At a 5-minute cadence the bot spends roughly $0.08 per cycle, $1.00 per hour, $24 per day, $720 per month. That's substantially cheaper than a single $500/month CoinAPI subscription, and you only pay when the bot is running. A bot that pauses overnight pays for nothing overnight.
Three reasons specifically for autonomous bots:
The same bearer token works on TensorFeed.ai. So if your trading bot also wants AI model intelligence (which model is currently best for code? for routing? for sentiment analysis?) or AI-news context, those calls draw from the same credit pool. One purchase, two data sources, two domains. Same wallet, same credits, same atomic-charge guarantees.
Buy 50 credits ($1 USDC) and you can run a trading bot for ~12 hours of 5-minute decision cycles. If your bot delivers ROI, top up. If it doesn't, you've spent a dollar.
Read the full agent-payments docs →
Other landing pages for builders: free API documentation, OpenAPI 3.1 contract, /llms.txt agent discovery file, /agents.txt capability manifest.