hypernatt-terminal
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@hypernatt-terminalGet the liquidation radar for Bitcoin and read the manifest first."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
HyperNatt Terminal
Forced-order map for AI agents + Li.Fi cross-chain swap. Pay-per-call via x402.
Most agents (and most humans) only see classic public indicators. This MCP shows where leveraged liquidations stack — a market-structure layer those indicators do not expose. Read-only context. Your agent still decides.
Built by one person. Code is public. Backend is private. No custody. Not trade advice.
Full HyperNatt platform (vault, assistant): https://hypernatt.com — this repo is one agent-facing brick.
Public mirror synced from a private monorepo. History here is mirror commits.
Quick start (30 seconds)
No install for the hosted MCP. In Claude → Settings → Connectors → Add custom connector:
Field | Value |
Name | HyperNatt Terminal |
URL |
|
Then ask:
Call get_agent_manifest, then get_liq_radar. If cluster_grammar is present, read it before liq_radar. class=noise (<3%) is high-leverage bait — ignore. class=true (~7%+) is the low-leverage stack. Terrain, not a signal.
More clients (Cursor / Cline / Codex / Windsurf): docs/integrations.md.
Related MCP server: Hyperliquid MCP Server
What HyperNatt Terminal gives your agent
A map of forced orders: where leveraged positions will be liquidated if price reaches them (clusters, OI build-up, observed liquidations).
This is market structure context, not a trade signal. It shows terrain so a sovereign trading agent can judge timing, sizing, and risk with more awareness than agents that only read RSI/MACD-style public feeds.
Your agent still decides. We show the terrain.
What your agent can do with this (honest scenarios)
Scenario 1 — Ignore noise; wait only near class=true
Ignore clusters with class=noise (distance < 3%). If cluster_grammar is present, wait only near class=true (low-leverage stack ~7%+). Then watch real_liquidations print — without assuming a guaranteed reversal. class=true is terrain, not an entry signal.
Scenario 2 — Size relative to distance
Cluster ~5% away vs ~0.5% away is a different risk parameter. Distance informs sizing or patience — it is not a prediction that the level will be hit.
Scenario 3 — After real liquidations, check OI
If price swept a zone and real liquidations spiked, compare with OI change. A large OI drop can mean forced flow is partly exhausted — still context, not an auto entry.
Glossary (short)
Term | Meaning |
Cluster | Zone where leveraged positions may liquidate if mark reaches it. Not a TP. |
| If present: read this first. |
Nearest vs largest |
|
Real vs estimated |
|
| Directional density bias from OI / L-S / funding. Not a hit probability. |
Live glossary + scenarios also ship on get_agent_manifest → agent_interpretation_rules_v1.
What you get
Tool | Price |
| Free — call first |
| $0.001 x402 (Base EIP-3009 or Solana SVM exact) |
| Free at MCP (you sign; gas + Li.Fi fee on-chain) |
Solana pay: do not reuse an EVM/Base x402 client. Use @x402/svm. See docs/x402-pay.md.
Whitelist get_liq_radar: BTC ETH SOL BNB XRP HYPE ZEC (omit symbol = BTC).
What we do NOT claim
We do not claim | What we do show |
Generic Coinglass wrapper | Operator-shaped forced-order map for agents |
Predictive Fuel Score / sweep classifier | Distance, size, OI, real liqs — labeled |
Custody of keys or funds | Agent signs own txs |
Trade advice / guaranteed edge | Read-only JSON context |
More than 3 MCP tools | Exactly 3 · v2.7.0 |
Vault / | Separate HyperNatt vault product |
Independent security audit | Public code + SECURITY.md |
Try it
curl -s https://hypernatt.com/api/m2m/agent/manifest
curl -i https://hypernatt.com/api/m2m/liq-radarStep | Link |
Quickstart | |
x402 Base vs Solana | |
Integrations (5 clients + AgentKit) | |
HL terrain example (read-only) | |
Honest interpret example | |
CDP Bazaar checklist | |
Examples | |
Security | |
Buyer skill (use the MCP) | |
Seller skill (x402 hardening) | |
Tests | test/ — |
Changelog |
MCP: https://hypernatt.com/mcp/protocol
For non-crypto users (x402 wallet)
If you do not already have a funded USDC wallet, install Coinbase Agentic Wallet MCP first (email/OTP + card onramp — no seed phrase):
npx @coinbase/payments-mcpDocs: Agentic Wallet MCP. Then connect HyperNatt Terminal (https://hypernatt.com/mcp/protocol). Your agent can pay get_liq_radar ($0.001) automatically when it sees HTTP 402.
Optional heavy use (not required): swap-earned quota or $5/mo Agent Pass — see live pass_program / quota_program on the manifest.
License
MIT — see LICENSE.
This server cannot be deployed
Maintenance
Related MCP Connectors
Agent-native crypto market-data over MCP+REST: order flow, whales, liquidations, calibrated scores
MCP server giving AI agents one-connection access to crypto & DeFi data: DeFi protocol TVL, stableco
Real-time Ethereum liquidation intelligence for AI agents and MEV bots.
MCP server connecting AI agents to non-custodial staking data across 130+ networks.
Related MCP Servers
- AlicenseAqualityDmaintenanceAn MCP server that provides AI agents with real-time access to DEX liquidity pool data, enabling smarter trading, analytics, and automated strategies.101MIT
- FlicenseNot gradedqualityCmaintenanceMCP server for Hyperliquid perpetual DEX with 48 tools for trading, monitoring, risk management, and security. Enables real-time market data, order management, risk validation, and automated trading strategies.-
- AlicenseNot gradedqualityDmaintenanceAn MCP server that provides DeFi data tools (crypto prices, whale concentration, funding rates) for AI agents via the x402 protocol.10 npm1MIT
- AlicenseNot gradedqualityBmaintenanceA set of MCP servers that provide AI agents with safe, composable access to web3 market data and trading, featuring read-only intelligence and execution modes with SIM/PAPER/LIVE safeguards.MIT