Data is Power.

The AI-native data layer for crypto. One API. One schema. No fragmentation. Watch an agent route, normalize, aggregate, and return confidence-scored data — live →

API Reference →

dataline-agent · live demolive

select a query below

⊘↵

[01] — pipeline

Intent in.
Confidence-scored data out.

Five stages, one API call. Each stage is testable, observable, replaceable independently.

Intent

intent.parse() — natural language to query

Describe what you need.

"BTC price" "ETH funding" "election odds"

Route

route.select() — best sources by latency

We pick the best sources.

smart routing multi provider fallback latency optimized

Normalize

schema.normalize() — same shape every venue

All data, one schema.

no vendor formats same shape across every venue

Aggregate

data_agreement() — cross source variance, outlier filtered

Reliable production output.

outlier filtering median / blended derived metrics

Output

output.emit() — JSON with confidence

Clean JSON with confidence.

{ symbol, price, confidence, source, ts }

[02] — coverage

Every market your agent might need.

Tier-A native integrations + Tier-B browser fallback (via GhostDriver) for the long tail.

◆

crypto markets

Spot · perps · orderbooks

→ Hyperliquid [live]
→ Coinbase
→ Binance
→ ...

pricevolumeorderbook

▲

prediction markets

Odds · events · resolution

→ Polymarket [live]
→ Kalshi
→ ...

oddseventsmarkets

✦

onchain data

Tx · balance · governance

→ Ethereum, Base
→ Solana
→ BNB · Sui · TON
→ ...

txbalancenft

★

web2 / social

News · sentiment · trends

→ Twitter / X
→ News APIs
→ Farcaster [soon]

newssentimenttrends

[03] — confidence

Your agent should know when to trust the data.

Every Dataline response ships with a confidence score so your agent can refuse or flag unreliable data BEFORE acting on it.

/ formula

confidence = data_agreement × source_reliability × freshness

data_agreement

Cross source variance, with outlier filtering.

source_reliability

Rolling uptime and latency per provider.

freshness

Time decay since each underlying source last ticked.

/ source breakdown · BTC price

Source	Latency	Reliability	Used
Binance	84ms	0.99	✓
Hyperliquid	92ms	0.97	✓
Coinbase	204ms	0.99	✓
CoinGecko	512ms	0.88	✗

Aggregated confidence99%

[04] — act layer

Read AND act on the same surface.

Same intent surface, write side. Non-custodial — your agent signs, your wallet broadcasts. Funds never leave the user.

trade

Hyperliquid trade-only

Open and close perp positions. Trade-only API permissions keep custody with your wallet.

swap

Uni v4 / Aerodrome

Token swaps with route quoting. We return signed tx; your wallet broadcasts.

bet

Polymarket bet

Place orders on prediction markets via the standard CLOB pattern.

browser

GhostDriver fallback

Long-tail dApps without an SDK — GhostDriver clicks through the UI for you.

[05] — playground

Try real data with natural language.

No signup. No credit card. Inspect the routing, the aggregation, and every raw source.

/ examples

/ settings

Show raw sources

Stream mode

dataline-playground · query● ready

> get btc price

Connecting to Dataline Hub... Authorizing... Routing query: 2 sources picked Paying $0.0001 USDC on Basechain... Aggregating responses...

{

"status": "success",

"data": {

"symbol": "BTC",

"price": 67123.45,

"currency": "USD",

"confidence": 0.98,

"source": "aggregated"

"timestamp": 1716283345

}

$0.0001 USDC · 87ms · 2 sources✓ cached for 30s

npm install·pip install·curl·MCP add-server

npm

pip

curl

MCP

# Install the SDK npm install @dataline/sdk # Make your first call import { Dataline } from '@dataline/sdk'; const tl = new Dataline({ apiKey: process.env.DATALINE_KEY }); const data = await tl.price('BTC'); console.log(data.price); // 67123.45 console.log(data.confidence); // 0.98

# Install the SDK pip install dataline # Make your first call from dataline import Dataline tl = Dataline(api_key=os.environ["DATALINE_KEY"]) data = tl.price("BTC") print(data.price) # 67123.45 print(data.confidence) # 0.98

# Hit the REST endpoint directly curl https://api.dataline.xyz/v1/price/BTC \ -H "Authorization: Bearer $DATALINE_KEY" # Response { "symbol": "BTC", "price": 67123.45, "confidence": 0.98, "source": "aggregated" }

# Add Dataline as an MCP server in Claude Desktop claude mcp add dataline \ --url https://mcp.dataline.xyz \ --header "Authorization: Bearer $DATALINE_KEY" # Or in any MCP-compatible host mcp install @dataline/server # Now ask: "what's the BTC price?" — agent calls Dataline directly