Elynd

CA: 0x5a6662b48b2bc42fc2ed0b1597f083f7bbf2630b

You can run several Elynd agents on one Hyperliquid account without them stepping on each other. Each agent gets its own capital allocation. It only spends that, and it only sells what it bought itself, so a DCA agent and a trend agent can share one account cleanly. A watchdog reconciles positions and balances against Hyperliquid and pauses an agent on any mismatch instead of trading on bad data.

Most trading bots ask you to trust them. Elynd agents don't need you to. Every trade an agent makes gets a receipt on Elysium: side, size, price, Hyperliquid order id. The track record is public and nobody can edit it, including us.

What gets checked before any order leaves an Elynd agent: - size vs max per trade and max per token - daily loss: past the cap, the agent stops buying and only protects positions - trades per day (protective stops don't count) - the market is on the agent's whitelist - liquidity: markets under $5K daily volume, spreads over 1.5% or orders above 2% of daily volume are rejected - price: IOC orders capped at 1% from mid Limits are stored in the AgentHub contract, not in a config file we can quietly change.

You don't have to trust an agent's PnL screenshot. You can check it yourself. Every trade emits a TradeRecorded event on Elysium: agent id, side, market, size, price and the Hyperliquid order id. Match the order id against Hyperliquid fills and you've verified the trade happened exactly as recorded. Every hour the agent also writes an EpochRecorded snapshot: equity and PnL since start. That's the track record the marketplace shows. Append-only, nobody can edit it, including us.

You don't need to know what an EMA crossover is to run a strategy. Tell our assistant what you want, like "slowly stack HYPE and BTC, buy more on dips, never sell". It picks the setup, fills in every setting and explains the risk. You check it, sign, and the agent starts.

How you know the agent you bought runs the strategy you were shown: The full strategy (preset, markets, params, limits, the instructions to the AI) is a JSON spec. Its keccak256 hash is stored on-chain with the agent as specHash. When you buy on the marketplace you get a locked copy: the spec can't be changed afterwards. Hash the spec you saw, compare it with getAgent(id).specHash, and you know it's the same thing that's running.

Getting started with Elynd from your AI assistant: 1. Sign in on elynd.trade, go to Account, API keys, create a key (ek_...). It's shown once and can be revoked any time. 2. Add the Elynd MCP server to Claude Code or Codex with that key. 3. Ask: "set up a low-risk DCA agent on HYPE and UBTC with 100 USDC". No key needed for public tools: presets, markets, models, the marketplace. elynd.trade/account?tab=keys

Buying a strategy usually means buying a black box. Not here. Before you buy an agent on Elynd you see everything: the strategy, the settings, the limits, the author's instructions to the AI. After the purchase your copy is frozen, so what you checked is exactly what runs.

Everything on elynd.trade runs on a public REST API, and you can use it directly. Public, no auth: - /v1/config: presets, markets, models, limits - /v1/marketplace: listings with on-chain track records - /v1/agents/:id/trades and /decisions: every trade and every AI decision with its reasoning With an API key you can validate strategies, prepare agents and manage your own. Full reference with curl examples is in the docs. elynd.trade/docs/mcp#key

Every decision an Elynd agent makes is public, not just the trades. GET /v1/agents/:id/decisions returns what the AI looked at, what it proposed and why, and which actions the risk engine rejected, with the exact rule that blocked them. Build a dashboard, a bot that tracks the best agents, or just check how an agent thinks before you buy it.

The Elynd MCP server never signs anything and never sees your keys. create_agent only prepares a draft and returns a link. You open it on elynd.trade and sign everything yourself with your own wallet: the trade-only key on Hyperliquid, the builder fee, one transaction on Elysium. Your AI assistant can design and manage agents. It can't move funds, by design.

Built an agent that actually works? Sell it. List it on the Elynd marketplace with its real on-chain track record. Buyers run their own copy on their own account, and you keep 98% of every sale.

What happens when the AI fails? If the model provider is down, times out or returns garbage, the agent doesn't guess. It holds and switches to protect-only mode: no new buys, while stop-loss, take-profit and trailing stops keep running in code. Stops never depend on the AI. They're checked by the runner on every cycle and fire on their own.

Elynd agents never hold your keys or your funds. Hyperliquid supports API wallets: a separate key you approve from your main account that can place orders but cannot withdraw or transfer. Each Elynd agent gets one, approved with a single signature (approveAgent). Worst case, if that key leaked, someone could place trades on your account. They could never move a single USDC out of it. And you can revoke it any time.

Elynd never holds your funds. Your agent trades from your own Hyperliquid account, inside the capital you give it. Several agents can share one account, each only spends its own part. Pause or stop any of them in one click.

How an Elynd agent actually places a trade: 1. Every few minutes the AI gets market data and its strategy, and proposes actions: buy, sell or hold, with a reason. 2. A deterministic risk engine checks each action against the agent's limits: size per trade, size per token, daily loss, trades per day, allowed markets, the live order book. 3. Anything that fails is dropped and logged with the reason. Only what passes is sent to Hyperliquid. 4. The fill gets a receipt on Elysium. The AI proposes. Code decides.

Before you start: you need USDC on your Hyperliquid spot balance (at least 10 USDC, around 100 is comfortable) and a little HYPE on Elysium for gas. Here is how to launch an AI trading agent on Elynd in about a minute: 1. Open elynd.trade, hit Create agent and connect your wallet 2. Tell the assistant what you want in plain words. It sets up the whole strategy 3. Check the safety limits and choose how much capital the agent gets 4. Sign twice on Hyperliquid: a trade-only key for the agent, it can never withdraw 5. Send one transaction on Elysium. Your agent is live It trades from your own account, every trade gets an on-chain receipt, and you can pause or stop it any time. Got a strategy that works? List your agent on the marketplace and keep 98% of every sale. elynd.trade

You can create an Elynd agent without opening our site, straight from Claude Code or Codex. The Elynd MCP server gives your AI assistant 13 tools: list_presets, list_markets, list_models, validate_strategy, create_agent, list_my_agents, get_agent_trades, stop_agent, list_marketplace and more. Describe what you want, the assistant builds the strategy spec, validates it against the preset rules and prepares the agent. elynd.trade/docs/mcp

AI is good at reading a market. It should never be the one deciding how much risk you take. That's why every Elynd agent has hard limits written into a smart contract on Elysium: max per trade, max per token, a daily loss cap, trades per day. Every action the AI proposes is checked against them before it's sent.

Why Elynd lives on Elysium, technically: Writing a receipt for every single trade only works if blockspace is fast and cheap. Elysium gives 100ms blocks and HYPE as gas, and it sits right next to Hyperliquid pricing and liquidity. On a slow or expensive chain you'd batch receipts or skip them, and "verifiable" would quietly turn into "trust us".