Model Context Protocol — Live on Hyperliquid

Your AI already knows how to trade.
Give it the data.

710K Hyperliquid wallets classified into behavioral cohorts — money_printers, whales, degens, giga_rekts. Ask Claude who's long, who's short, where liquidations cluster. Get answers no dashboard shows.

26
MCP Tools
710K
Wallets Classified
1.8B
Trades Indexed
5
PnL Tiers
Get Started — Free Browse All 26 Tools
What your AI can do
26 tools across 5 categories

Every tool returns structured JSON. One natural language question → the right tool fires automatically.

5 tools

Core Tools

"Profile this wallet." Full PnL tier, size tier, win rate, profit factor, and trade history in one call. The starting point for every investigation.

5 tools

Trader Analytics

"Who are the hidden gems?" Find traders with 70%+ win rate that no leaderboard surfaces. Compare any two wallets side by side.

6 tools

Market Intelligence

"Where do liquidations cluster?" Live funding rates, long/short ratios, OI by tier, and price-level liquidation zones updated continuously.

6 tools

Real-Time Trade Flow

"What did whales just buy?" Stream fills from leviathan wallets, track position opens/closes, and measure per-minute trade velocity.

Only Here 4 tools

Cohort Intelligence

"Are money_printers long or short ETH?" Behavioral tier consensus, tier migrations, and contrarian giga_rekt signals. Data no one else computes.

Complete API Reference
Every tool, every parameter

Your AI auto-selects the right tool. But if you want to know exactly what's available — here's the full catalog.

Core Tools

5 tools
pulse_trader_profile(wallet)

Full behavioral profile — PnL tier, size tier, win rate, profit factor, and trade statistics.

live_cohort_positions(pnl_tier, coin)

Current open positions for wallets in a specific PnL tier holding a given coin.

top_traders(metric, limit=25)

Top N traders ranked by pnl, win_rate, volume, profit_factor, or trade_count.

live_cohort_bias(coin)

Per-tier long vs short counts and notional sums from open positions.

recent_large_trades(min_usd=100k, limit=50)

Fills above a notional threshold in the last 24 hours, newest first.

Trader Analytics

5 tools
rank_traders_by_risk_adjusted(limit=25)

Ranked by risk-adjusted score: win_rate × profit_factor.

find_hidden_gems(min_win_rate=60, min_trades=100, limit=25)

Skilled but unknown traders: high win rate, 100+ trades, not in money_printer tier.

wallet_trade_history(wallet, coin?, limit=50)

Recent fills for a wallet, optionally filtered by coin.

cohort_summary()

Overview of the entire classified wallet universe — tier distribution and median stats.

compare_wallets(wallet_a, wallet_b)

Side-by-side stats comparison of two wallets.

Market Intelligence

6 tools
long_short_ratio(coin)

Total longs vs shorts and summed notional in the positions table.

top_coins_by_volume(hours=24, limit=20)

Coins ranked by total traded notional in a given time window.

funding_rate_snapshot()

Live funding rates for all Hyperliquid perps from the REST API.

liquidation_zones(coin, bucket_usd=100)

Price levels where open positions cluster, signaling potential liquidation cascades.

open_interest_by_tier(coin)

Open interest breakdown for a coin by trader PnL tier.

market_activity_pulse(coin)

Trade count and volume over 1-minute, 5-minute, and 1-hour windows.

Real-Time Trade Flow

6 tools
largest_trades_today(coin?, limit=20)

Top N fills by notional in the last 24 hours, optionally filtered by coin.

whale_activity_feed(limit=20)

Recent fills from leviathan and whale-tier wallets in the last hour.

position_opened(coin, limit=20)

Recent position opens (Long / Short) sorted by notional size.

position_closed(coin, limit=20)

Recent position closes sorted by absolute realized PnL.

smart_money_flow(coin?)

Net notional flow (buys − sells) from money_printer wallets over 1h, 4h, and 24h.

trade_velocity(coin)

Trades per minute for a coin over the last 10 minutes.

Unique Cohort Tools

4 tools — only here
money_printer_holdings(coin?, limit=30)

Current open positions held by money_printer-tier wallets — the best traders on Hyperliquid.

giga_rekt_activity(coin?, limit=30)

Recent trades by giga_rekt-tier wallets — biggest losers, useful as a contrarian signal.

tier_migration(days=7, limit=30)

Wallets that changed PnL tier recently — rising stars or falling whales.

cohort_consensus(coin)

Per-tier percentage of wallets long vs short — institutional sentiment proxy.

Why This Server
What you can't get anywhere else

Other data tools show you charts. This one lets your AI reason about Hyperliquid in real time.

🧠

Behavioral Wallet Cohorts

710K wallets scored by realized PnL and classified into money_printer, profitable, breakeven, losing, and giga_rekt. Not a leaderboard snapshot — a continuously updated behavioral map.

✓ We compute it  ·  ✗ CoinGecko / DeFiLlama don't
🎯

Cohort Consensus

Ask "Are money_printers long or short ETH?" and get an instant answer with notional sums. The only API that exposes per-tier directional bias on any coin.

✓ Per-tier long/short breakdown  ·  ✗ Others show aggregate only
⚡

AI-Native via MCP

Your AI doesn't call a REST API — it calls MCP tools natively. No HTTP client, no auth headers, no response parsing. Ask a question, get a structured answer.

✓ Claude calls tools directly  ·  ✗ REST APIs need glue code
How it works
We index the clearinghouse. Not a third-party API.

Most "Hyperliquid data" products relay CoinGecko or DeFiLlama. We capture every fill directly from the Hyperliquid L1 and compute analytics that don't exist anywhere else.

710K

Wallets → Behavioral Cohorts

Every wallet scored on realized PnL, trade count, and volume. Classified into 5 PnL tiers (money_printer, profitable, breakeven, losing, giga_rekt) × 5 size tiers. Re-computed continuously.

Real-Time WebSocket Ingestion

Fills stream in via WebSocket, not scraped hourly. Position opens, closes, and liquidations hit TimescaleDB within seconds. Your AI sees what just happened, not what happened an hour ago.

1.8B

Full Trade Lifecycle

Every trade stored with entry price, exit price, hold duration, side, leverage, and realized PnL. Continuous aggregates pre-compute volume, velocity, and flow windows (1m, 5m, 1h, 24h).

FastMCP + asyncpg

Python MCP server using FastMCP over stdio. Connection-pooled asyncpg queries. Sub-100ms response times. Drop the JSON config into Claude Desktop and start asking questions.

Output Format

Ask “Who is this wallet?”

One call to pulse_trader_profile returns everything your AI needs to reason about a trader — tier, win rate, leverage, hold time, PnL.

No parsing HTML. No scraping leaderboards. Structured JSON that Claude, GPT, or your own agent can immediately analyze, compare, and act on.

JSON Response
{ "address": "0x7fda...7d1", "total_pnl": 847231.45, "trade_count": 2847, "win_rate": 68.4, "total_volume": 142500000, "largest_win": 124500.00, "largest_loss": -31200.00, "avg_leverage": 4.2, "avg_hold_minutes": 340, "pnl_tier": "money_printer", "size_tier": "whale", "profit_factor": 2.87 }
In Action
Natural language → institutional data

You ask a question. Claude picks the right tool. You get data that would take 30 minutes to assemble manually.

claude — nexxore mcp connected
You
Are the best traders long or short ETH?
Claude
▸ live_cohort_bias("ETH")
money_printer wallets are decisively short — 85 shorts vs 28 longs, $287M short notional. Leviathan tier is the only group with long bias (58%). The smart money consensus is bearish ETH.
You
Show me where ETH longs would get liquidated
Claude
▸ liquidation_zones("ETH", bucket_usd=50)
Major long liquidation cluster at $1,820–$1,870 — $42M notional from 127 positions. Secondary cluster at $1,750 ($18M). Below $1,700, thin. If ETH drops 4% from here, expect a cascade.

Questions your AI can answer now

What are money_printers holding right now?
Net smart money flow on BTC last 4 hours
Who are the hidden gem traders?
Compare 0xabc... vs 0xdef... head to head
Which wallets changed PnL tier this week?
Top 10 coins by notional volume today
30 Seconds to First Query
Connect your AI to Hyperliquid

No cloning repos. No Docker. Paste one JSON block and start asking questions.

1

Get your API key

Free tier. No credit card. Takes 10 seconds.

2

Drop in the config JSON

One JSON block in Claude Desktop, Cursor, Claude Code, Cline, or any MCP client.

3

Ask your first question

"What are money_printers holding?" — your AI calls the right tool and returns structured data.

claude_desktop_config.json
{ "mcpServers": { "coinversaa": { "command": "npx", "args": ["-y", "@coinversaa/mcp-server"], "env": { "COINVERSAA_API_KEY": "your-api-key" } } } }
Common Questions
FAQ
Three things no one else does: (1) We classify 710K wallets into behavioral cohorts based on realized PnL — money_printer, profitable, breakeven, losing, giga_rekt. (2) We expose live cohort positions — what the best traders are holding right now. (3) We deliver it via MCP, so your AI can query it natively without you writing a single API call.
MCP (Model Context Protocol) is Anthropic's open standard for connecting AI models to external tools and data. Instead of copy-pasting data into prompts, your AI calls tools directly — like smart_money_flow("BTC") — and gets structured JSON back. It turns Claude from a chatbot into a trading terminal.
We classify wallets into 5 tiers based on total realized PnL: money_printer (top performers), profitable, breakeven, losing, and giga_rekt (bottom). Size tiers (leviathan, whale, dolphin, shrimp) are computed from total traded volume. Both are re-computed continuously as new trades are indexed.
Any MCP-compatible client: Claude Desktop, Cursor, Claude Code, Cline, Continue, Zed, Windsurf, and custom MCP clients. The server runs via npx @coinversaa/mcp-server — paste the JSON config and it works immediately. No cloning, no Docker, no building.
Fills are indexed via WebSocket in real time — seconds of latency, not hours. Cohort classifications are re-computed on a continuous loop. Funding rates are fetched live from the Hyperliquid REST API on every call. When you ask "what happened in the last 10 minutes," you get the last 10 minutes.
Yes. The free tier is generous enough for personal use, research, and prototyping agent workflows. Paid plans unlock higher rate limits for production agents, fund-level analytics, and commercial applications.

Give your AI the data edge

26 tools. 710K classified wallets. Cohort consensus, smart money flow, liquidation zones. One JSON config away.

Get Started — Free Browse All 26 Tools GitHub