# HAIPA Documentation — Full Text > Complete HAIPA documentation for AI agents, concatenated into one file. HAIPA is non-custodial, institutional-grade trading infrastructure for AI agents on Hyperliquid. Operated by Hekaton LLC. Source of truth: https://haipa.ai/docs --- # What is HAIPA? HAIPA is non-custodial institutional-grade trading infrastructure for agents on Hyperliquid, built on Tokyo-colocated compute and persistent virtual machines equipped with a suite of institutional-grade trading execution algos. You define the risk caps for your agent: size limits, leverage caps, position boundaries. HAIPA enforces them at the signing boundary before every order fires, so your agent gets institutional speed and capabilities without ever breaking your rules. This guide takes you from a fresh wallet to a live, policy-bounded trading agent in a few steps. ## Overview HAIPA is a non-custodial policy engine for agents and bots on Hyperliquid that bounds every trade before it fires. It is backed by real institutional-grade execution infrastructure: Tokyo-colocated compute, persistent virtual machines, and algo primitives such as trailing stop and peg orders — live now, with more shipping over time. Your agent is armed with the same tools institutional HFT desks have run for decades, but never outside your rules. ## Who is this for? HAIPA is built for two kinds of users — the difference is who's in the driver's seat. 1. **You want a junior trader.** Talk to trade. You're in the driver's seat: just describe what you want and your agent executes, spawning algos, placing trigger orders, making markets on your behalf. It keeps working after you close the chat window, and it works inside rules you set. 2. **You want a trading framework for your persistent agent.** Your agent already runs 24/7, but it's flying blind with no guardrails. Drop in our MCP connector and it can launch algos on the fly, implement custom strategies, and trade within policy you control. Your agent's harness might run on a Mac mini in your office, but its execution engine lives at the exchange in Tokyo. Either way, every action your agent takes still clears your policy before it fires; the rules are always yours. --- # Quick Start Already have a funded Hyperliquid account and a wallet like MetaMask connected? This is the fastest path to trading with HAIPA. The only thing you skip versus a first-time setup is **funding** — your account already holds USDC, so you can go straight to provisioning your agent. (Need to add funds first? Start with Bridge Funds, then come back here.) See the full sections for details on any step. 1. **Connect your wallet**, if you haven't already. → Connect MetaMask wallet 2. **Generate your agent key and set your policy.** Key generation and policy setup both happen on the Agent Rails dashboard. Set your **Markets** and **Risk** limits now; **Algos** and **Treasury** can wait until you need them. → Generate a HAIPA Agent key · Configure your policy 3. **Connect your LLM.** Link Claude or ChatGPT over MCP so you can start describing trades in plain English. → Connect your LLM 4. **Trade.** Every order your agent proposes is still checked against your policy before it's signed. That's it — HAIPA is live. --- # Getting Started ## Connect your MetaMask wallet **Purpose:** Connect your MetaMask wallet to HAIPA so you can begin configuring your trading agent. **Steps** 1. Click **Connect Wallet** on the HAIPA landing page. 2. Select **MetaMask** from the list of supported wallets. 3. When MetaMask opens, choose the specific account you want to use. 4. Review the connection request. 5. Click **Connect** to attach the wallet. 6. Click **Sign In** and sign the SIWE request in your wallet. This signature is only used to prove your ownership of the account and doesn't move funds. 7. Once the sign-in completes, HAIPA takes you to the Agent Rails dashboard, where you can begin configuring your agent's risk policy. **Note: Non-custodial** Connecting your wallet does not give HAIPA custody of your assets. HAIPA is a non-custodial policy engine. Every action your trading agent takes must still satisfy the trading policies you define before any order is submitted. ## Generate a HAIPA Agent key This is the short version to keep the setup flow moving. The full walkthrough — including rotation and the Reset tab — is in Understanding the HAIPA Key System. You can provision an agent key during onboarding, or later after resetting your existing agent key: 1. In the HAIPA admin UI (**app.haipa.ai**), open the **Provision** tab and click **Provision Agent**. 2. Approve the agent key by signing an **ApproveAgent** transaction on Hyperliquid with your wallet. On the **Authorize** page, click **Approve Agent**. HAIPA agent keys are named either **haipa-mcp** or **haipa-algo**. Disabling either agent in the official Hyperliquid UI is your prerogative, but doing so severs that agent's ability to manage orders on your behalf and requires resetting the agent key in the HAIPA admin (**app.haipa.ai**). --- # Understanding the HAIPA Key System Rotating your Agent, API, and Algo keys. **Note: The three keys at a glance {#keys-glossary}** - **API Key (K)** — the key you give your LLM chat agent; it authenticates with the HAIPA MCP server. Your agent knows this key but never knows the underlying Agent Key. - **Chat Agent Key** — the on-chain agent key HAIPA uses to place your MCP-driven trades and launch algo orders. Kept separate from your chat agent so guardrails can be enforced before anything reaches Hyperliquid. - **Algo Agent Key** — a dedicated signing key that runs your persistent autonomous algos (trailing stop, peg) server-side in Tokyo. ## What is a HAIPA Agent key? It's an agent key you approve on Hyperliquid that lets your agent trade on your behalf. This key cannot withdraw funds. HAIPA also keeps this key separate from your agent, so we can enforce your risk guardrails before any transaction is broadcast to the Hyperliquid network. ## How to generate a HAIPA Agent key You can provision an agent key during onboarding, or later after resetting your existing agent key: 1. In the HAIPA admin UI (**app.haipa.ai**), open the **Provision** tab and click **Provision Agent**. 2. Approve the agent key by signing an **ApproveAgent** transaction on Hyperliquid with your wallet. On the **Authorize** page, click **Approve Agent**. HAIPA agent keys are named either **haipa-mcp** or **haipa-algo**. Disabling either agent in the official Hyperliquid UI is your prerogative, but doing so severs that agent's ability to manage orders on your behalf and requires resetting the agent key in the HAIPA admin (**app.haipa.ai**). ## When and how often should you rotate your agent key? By default, agent keys expire three months after creation. You can rotate them as often as you like. You can see the expiry for each key in the admin UI (**app.haipa.ai**). Your Chat Agent Key expiry appears at the top of the screen. If you've enabled algo trading, you'll also find your Algo Agent Key expiry on the Algos tab. You can rotate either key in the **Reset** tab of the HAIPA admin UI: Rotating the **Algo Agent Key** asks you to sign for your new agent key in your wallet. This causes a temporary pause in your running agents until they detect the rotation (usually a few seconds). You can verify your algos are still running by asking your HAIPA MCP Chat Agent to *"list my algos."* Rotating the **Chat Agent Key** deprovisions your old key immediately, and you then re-enter the normal onboarding flow — except that your risk guardrails are pre-populated from your previous profile. You'll be asked to sign to approve your new agent key, then you'll receive a new MCP URL to give your chat agent. Chat Agent Key rotation also pauses all running algos, which you can resume by asking your HAIPA MCP Chat Agent to *"resume my algo orders."* ## The Reset tab This tab can look confusing because there are so many keys. What's the difference between rotating your API Key (which we'll call **K**), rotating your Algo Agent Key, and deleting your agent? Here's how each one works. Your API Key is what you give to your LLM Chat Agent; it's how the agent authenticates with the HAIPA MCP server. HAIPA sits between your LLM agent (the "Chat Agent") and the actual Agent Key. Your agent knows the API Key but, critically, never knows the actual Agent Key — that separation is how we can enforce your guardrails. Deleting an agent removes the agent's key and revokes its access to HAIPA. It does **not** delete the agent itself or affect the assets in your Hyperliquid account. --- # Bridge Funds via Across Integration Our admin UI (**app.haipa.ai**) supports cross-chain transfers of USDC collateral from Solana and EVM chains such as Ethereum and Arbitrum to Hyperliquid (Hypercore) through our Across integration. If you've never funded your Hyperliquid account, you'll be prompted to deposit via Across during initial onboarding. Afterwards, you can reach the bridging UI from the Treasury tab (see screenshot 1 below). 1. **Select a chain** — bridge from Solana or Ethereum. 2. **Account selection** — this defaults to your Hyperliquid account, but you can also fund another Hyperliquid address by selecting **Another Address**. 3. **Get Quote** — once you've chosen a source account, deposit amount, and deposit address, click to pull a current quote for the bridge (technically a cross-chain swap) from the Across network of relayers (see screenshots 2 and 3 below). ## Deposit: selecting the source account **Paying Wallet.** By default, the source wallet for the deposit is set to your logged-in user. To use a different wallet, click the wallet tile (1), select **Disconnect**, then connect the desired source account in your wallet extension. Your destination account stays connected as well. ## Get Quote 1. **Input Amount** — the amount that will leave your deposit address, i.e. the bridging swap input. 2. **Recipient** — another chance to double-check you're sending to the right Hyperliquid (Hypercore) account. 3. **Output Amount** — the input amount minus Across relayer fees (right now, HAIPA takes 0 fees on bridge transactions). 4. **Estimated Time of Arrival** — estimated from historical results; the actual time depends on network conditions and may be shorter or longer. 5. **Deposit Button** — initiate the deposit, then sign in your wallet to approve the deposit transaction on the source network. --- # Configure Your Policy Your policy is the set of rules your agent must satisfy before any order is signed. You set it across four tabs in the admin UI: **Markets**, **Risk**, **Treasury**, and **Algos**. ## The Markets tab The Markets tab is where you tell your agent which perp instruments it can trade. You can also set max leverage per market and set maximum and minimum position guidance. The numbers below match the callouts in the screenshot: 1. **Market** — the canonical Hyperliquid symbol for that product, e.g. `xyz:SP500`. 2. **Max Leverage** — set the max leverage to a custom value lower than the market's native maximum to manage risk. For example, limit `xyz:SP500` to 20x instead of the 50x the on-chain perp market supports natively. 3. **Min Size\*** — the minimum net position you want to accumulate. For example, if you don't want to short more than 20 perp contracts, set Min Size to `-20`. 4. **Max Size\*** — the maximum net position you want to accumulate. For example, if you don't want to go long more than 100 perp contracts, set Max Size to `100`. 5. **Remove** — for your agent to trade a perp market, it must be in this list. Removing a market bans your agent from placing new orders in that instrument, even if you still hold an open net position. 6. **Add Market** — search for markets by symbol. 7. **Save Markets** — you must press this to persist your changes. Size bounds are signed coin amounts (negative = short side). Leaving a bound blank means unbounded; a **min of 0** forbids shorting and a **max of 0** forbids longing. **Note: Important** \* Enforced best-effort pre-trade. Under certain on-chain race conditions, your agent may briefly exceed these bounds. We therefore recommend also setting Max Leverage, which produces an effective max and min position that is enforced on-chain (post-trade) in addition to HAIPA's pre-trade checks. ## The Risk tab The Risk tab is where you tell your agent how to calculate your account's PnL. You can set a cut-off PnL: below that level, your agent is denied from placing any further orders until you manually intervene — either by taking over trading or by re-adjusting your risk limits. The numbers below match the callouts in the screenshot: 1. **PnL Floor (USD)\*** — the minimum PnL you're willing to accept before your agent is cut off from trading. This is the level at which you want to intervene and stop a runaway agent. 2. **Starting Equity Baseline (USD)** — the amount of USDC you started with. HAIPA needs this to calculate current PnL as `Current Equity − Starting Equity`. 3. **Save Risk** — press this to persist your risk settings. 4. **Reset Baseline to Live Equity** — resets your Starting Equity to your current equity, effectively resetting your PnL to zero. **Note: Important** \* Enforced best-effort pre-trade. Your account can still lose more than this; the PnL floor is a mechanism for locking out your agent during a losing streak so you can manually intervene. ## The Algos tab The Algos tab is where you enable, monitor, and manage your agent's autonomous algo orders. The numbers below match the callouts in the screenshot: 1. **Status** — whether algo trading is currently enabled. 2. **Credit** — prepaid credits not yet debited. 3. **Subscription** — when the next subscription payment is due. 4. **Key** — your Algo Agent Key. 5. **Key Expiration** — when your Algo Agent Key expires and needs rotation (see When and how often should you rotate your agent key?). 6. **Subscription & Signing Key Details** — expandable details for your algo subscription and signing key. Server-side algo execution runs on a dedicated signing key with a prepaid subscription (30 USDC / 30 days). 7. **Working orders** — all your currently working algo orders and their basic information. 8. **History** — your previous algo orders that have filled or been canceled. Expanding **Subscription & signing-key details** reveals the **Disable Algo Key (kill switch)**: Disabling the Algo Agent Key here disables **all** running algos. This is the nuclear option. If you only want to rotate the key, use the key rotation steps instead. Your working and historical algo orders appear in the Working orders and History tables: ## The Treasury tab The Treasury tab is where you see what your account is worth and where you deposit and withdraw. The numbers below match the callouts in the screenshot: 1. **Balances** — shows total account value and the breakdown of value and required collateral across spot and the DEXs (e.g. `xyz`). 2. **Deposit from another chain** — using the Across Bridge. 3. **Withdraw to Arbitrum** — the native withdraw method on Hyperliquid. It costs 1 USDC per withdrawal (this is Hyperliquid's native base fee, not charged by HAIPA). 4. **Native USDC send** — send USDC to an external account. --- # Connect Your LLM Connect your chatbot to HAIPA over MCP so you can describe trades in plain English. Whichever assistant you use, it only ever proposes trades — HAIPA checks every one against your policy before it's signed, and never takes custody of your funds. ## Connect Claude (MCP) **Purpose:** Connect Claude to HAIPA over MCP so you can describe trades in plain English and have Claude execute them within your policy. **Steps** 1. Copy your HAIPA MCP URL from the Agent Rails dashboard (see Generate a HAIPA Agent key if you haven't created one yet). 2. In Claude, open the **Customize** menu, then **Connectors**. 3. Click **Add**, then select **Add custom connector**. 4. Paste your HAIPA MCP URL as the remote MCP server URL. 5. Click **Add** to finish, then confirm the connection. Claude should now list HAIPA as an available tool. 6. Test it with a simple request, like asking Claude to check your current positions or get your account info. Claude only ever proposes trades. HAIPA checks every one against your policy before it's signed. If a request breaks your rules (for example, exceeding a size or leverage cap), HAIPA blocks it and Claude will typically offer a compliant alternative. ## Connect ChatGPT (MCP) **Purpose:** Connect ChatGPT to HAIPA over MCP so your chatbot can propose trades that are checked against your policy before they execute. **Steps** 1. Copy your HAIPA MCP Server URL from the Agent Rails dashboard. 2. In ChatGPT, open **Settings**. 3. Select **Plugins**. 4. Scroll down and click **Browse plugins**. 5. On the Plugins page, click the **+** button in the upper-right corner. 6. In the **New Plugin** dialog, enter: - **Name:** HAIPA - **Description (optional):** AI trading assistant for Hyperliquid - **Connection:** leave **Server URL** selected - **Server URL:** paste your HAIPA MCP Server URL - **Authentication:** select **No OAuth** — a HAIPA account does not require OAuth. 7. Review the security warning and check **"I understand and want to continue."** 8. Click **Create**. Once connected, HAIPA appears in your installed plugins and is ready to use in ChatGPT. As with Claude, ChatGPT never bypasses your policy. Every order it proposes is checked by HAIPA before it's signed, and HAIPA never takes custody of your funds — regardless of which chatbot you connect. --- # Trading Tools for Your Agent ## Algos overview HAIPA pairs its policy engine with institutional-grade infrastructure for your agent: Tokyo-colocated compute, persistent VMs, and algo primitives. **Trailing stop**, **peg (chase)**, and **trigger bracket** orders are live now, with more algo primitives shipping over time. Your agent gets the same tools institutional HFT desks have run for decades, but every order is still checked against your rules before it fires. Non-custodial, always. ## Trailing stop order A trailing stop is an exit order that follows the market instead of sitting at a fixed price. As your position moves in your favor, the stop trails behind it at a set distance. If the market reverses and pulls back by that distance, the order fires and closes the position — locking in gains without capping how far the upside can run. It protects you when the position goes against you, like a static stop. But it also protects profits when the position moves your way, because the stop is dragged along as the market moves in your favor. For example: *"buy 10 BTC and place a trailing stop 800 points lower."* On HAIPA, the trailing stop is one of the algo primitives your agent can use as part of its execution toolkit. But it doesn't sit outside your rules: every trailing stop your agent places is still checked against your policy caps (size, leverage, position limits) before it's submitted. HAIPA never holds your funds and never overrides what you've set — it makes sure the order matches your rules before it goes out. **Note: Important — read this before placing one** HAIPA currently supports **one trailing stop per instrument**. You can't place two trailing stops on the same instrument with different offsets (for example, one at offset 150 and another at offset 200 on the S&P 500). It's one or the other, and it applies to the full position. If you or your agent gets flat some other way, or flips the position, the trailing stop auto-cancels. The **offset is set in absolute price points, not a percentage.** For example: - On the S&P 500, the offset is in index points. - On SOL or BTC, it's in USD per coin. - On Brent oil, it's in USD per barrel, and so on. You can see your running trailing stops in the admin UI under the **Algos** tab at **app.haipa.ai**. You can also ask your agent to *"list my algos"* to list all running algo orders, and you can pause, resume, or cancel any of them. - **Pause** freezes the stop at its current level — it stops trailing but stays in place, so you keep your protection. **Example prompts** - *"Place a trailing stop in xyz:sp500, offset 100."* - *"Trail a stop 40 points above my xyz:TSLA short."* - *"What are my options for protecting my position?"* ## Peg (chase) order A peg order (chase order) is a limit order that automatically reprices itself to follow the best bid or ask as the market moves, so it keeps resting at the top of the book instead of getting left behind. The order moves up (or down) toward your limit price as the market moves; once it reaches your limit, it rests there as a normal limit order. You can also configure a peg order to act like an iceberg by setting the **Max Show** quantity lower than the order's total quantity. An iceberg order displays only a small portion of the total size (the "tip") on the order book at a time, automatically replenishing the visible slice as each piece fills, so the full size stays hidden from other traders. Post-only (maker) orders pay lower fees natively on Hyperliquid; aggressive (taker) orders pay a higher fee to the exchange. You can ask your agent to *"list my algos"* to list all running algo orders, and you can pause, resume, or cancel a running peg order. **Example prompts** - *"Buy 100 xyz:BRENTOIL perps using a peg order, max show 10 lots at a time, limit 94."* - *"Sell 10 xyz:sp500, peg order, max show 2 perps, limit 7450."* - *"Use a peg order to buy 50 HYPE, limit 60."* ## Trigger bracket order {#trigger-bracket} A trigger bracket is a fully self-managing breakout trade: you set a trigger price, and if the market reaches it, the algo enters at market and immediately places a stop loss to protect your entry (and optionally a take-profit to exit a winner) at your desired offset. If the trade works, the take-profit closes it. If it doesn't, the stop closes it. Either way, no follow-up action is needed from you or your agent. The bracket manages both the happy path and the unhappy path on its own. For example: *"If the SP500 breaks 7465, buy 10, stop 65 points below the trigger, take profit 130 above."* Sells work the same way: the trigger fires when the market falls to your price, with the stop above and the target below. ### How trigger brackets work The trigger fires on Hyperliquid's mark price, exchange-side, so your entry triggers even if you're offline. The stop and take-profit are sized to what actually filled (never more), and both are reduce-only: a trigger bracket can only open or add to a position, and its exits can only reduce it. Over a full round trip, a bracket's net effect on your position is zero. Because of this, a bracket that would trade against your existing position is rejected upfront. If you're short SOL, a sell bracket is fine, but a buy bracket would reduce your short rather than open a trade, and that's a job for a regular order. ### Stacking brackets You can stack up to ten brackets in a single perp market, all inside one algo. The boundary is one trigger-bracket algo per market, with up to ten brackets inside it. Stack same-direction brackets as an entry ladder, or place a buy above and a sell below to trade a breakout in either direction. If the two ever cross, they simply close each other out, which is reflected in the algo's stats. **Note: Offset and target parameters** The `offset` and `target_offset` parameters are in absolute price points, using the same convention as trailing stops: - On the S&P 500, the offset is in index points. - On SOL or BTC, it's in USD per coin. - On Brent oil, it's in USD per barrel. Both are measured from the trigger price (stop = trigger − offset, target = trigger + offset for buys), and each must be at least 0.10% (10 basis points) of the trigger price. The take-profit rests as a limit order, so it typically earns the cheaper maker fee. Its contract is price-or-better: if the market has already run past your target by the time it's placed, it fills immediately. ### Adjusting a running trigger bracket You can add brackets to a running algo or edit them live: *"add a third bracket at 7500,"* or *"move the stop on my fired bracket to 40 points."* Running brackets are never disturbed by edits to their neighbors. Adding brackets re-checks your account's risk rails at the new combined size. ### Managing your trigger brackets You can see your running trigger brackets in the admin UI under the **Algos** tab at **app.haipa.ai**, with per-bracket state (armed / fired / protected / done), fills, and notices. You can also ask your agent to *"list my algos"* and it will list all your running algo orders. You can pause, resume, and cancel a running trigger bracket. - **Pause** pulls the entry triggers (no new positions can open while paused) but leaves the stops and take-profits of any bracket that already fired in place, so protection stays on. - **Resume** re-arms the entries, unless the market crossed a trigger while you were paused (re-placing it would fire instantly, so the algo tells you instead). **Note: Caution on cancel** Cancelling the algo pulls everything it owns, including the stop and take-profit of a bracket that already fired. The position that bracket was managing is left unprotected. The cancel response warns you when this is about to happen. Re-protect the position or close it if that's not what you intended. **Example prompts** - *"Place a trigger bracket on xyz:sp500 — buy 10 if it breaks 7465, stop 65 points below the trigger, take profit 130 above."* - *"Bracket xyz:BRENTOIL both ways: buy 5 if it breaks 89, sell 5 if it breaks 87, stops 50 cents, targets 1 dollar."* - *"Add a third bracket to my sp500 algo: buy 5 more if it breaks 7500, same offsets."* - *"Move the take-profit on my fired bracket to 40 points."* - *"Ladder into HYPE: buy 20 at 62, 20 more at 64, 20 more at 66, each with a 1 dollar stop."* --- # HAIPA Prompt Cookbook A quick reference for the kinds of requests you can give your agent. Every recipe below has its own shareable link — copy the section link to point someone straight at it. ## Get your account Your agent can get your account information, including equity, margin requirements, position information, and liquidation prices. **Example prompts** - *"Use HAIPA to get my account."* - *"Check my account after every successful fill."* **Note: Pro tip** Include the instruction *"Check my account after every successful fill"* in your opening prompt. Your agent will then review the account after each transaction to confirm it's in the expected state. ## Place orders on Hyperliquid Place a market or limit order in a specific perp market. **Example prompts** - *"Buy 100 HYPE at market."* - *"Sell 10 xyz:sp500 limit 7555."* - *"Get me a quote for selling 10 xyz100 and execute if slippage is less than 10 bps."* ## Place a trigger / set Take Profit / Stop Loss (TPSL) Place a trigger order to buy or sell a perp market. Alternatively, use Set TPSL to attach a stop loss or take profit to an existing position. **Example prompts** - *"Buy 10 sp500, then attach a bracket: profit target at 7777, stop loss at 7444."* - *"Place a stop-loss order to sell my SOL position at 65, reduce-only."* - *"Attach a stop-loss order to my sp500 short, buy trigger 7605."* - *"I want to be long, but only if we rally first — set a buy trigger for 350 in xyz:TSLA."* ## Cancel orders Cancel a single order by ID, or cancel all orders in a perp market. **Example prompts** - *"Cancel my stop-loss order in HYPE."* - *"Cancel all my orders in xyz100."* - *"Look up the order ID of my top bid, cancel it, and replace it 10 points lower."* ## Close a position Exit your entire position in a single perp market. **Example prompts** - *"Dump my xyz:GOOGL."* - *"Get flat xyz100."* ## Get open orders Query your native Hyperliquid orders working on the exchange, and whether or not they're attached to algo orders. Useful for syncing your agent's state with the state of your order book. **Example prompts** - *"Get my open orders and cancel all the buys."* - *"What orders are open right now?"* ## List algos Query all your algos and their current status (running or paused). Useful for checking the live state of an algo — for example, where your trailing stops currently sit, or how much of your iceberg order has filled. **Example prompts** - *"List all my algos — are they all running?"* - *"What's the status of my peg order in sp500?"* - *"Which algo orders are paused?"* **Note: Algo controls** Once an algo is running, you can ask your agent to pause, resume, cancel, or update it. See Algo controls below for what each one does. ## Algo controls Every algo order shares the same control surface — pause, resume, cancel, and update. This is a general guide; consult the docs for each algo order type to see exactly how that algorithm responds to each control. - **Pause** — stops updating the algo (no new orders). Exact behavior varies per algo, but in general: protective orders that bring your net position closer to zero (e.g. a stop loss) keep working, while orders that increase the magnitude of your net position may be pulled. - **Resume** — the algo resyncs its working orders in the market with the desired working-order state, then proceeds normally from there. - **Cancel** — stops updating and removes all working orders from the book. - **Update** — modify algo parameters on the fly. For example: *"tighten up my trailing stop — move the offset from 100 to 50."* ## Get docs A tool that helps your agent answer your questions — and its own — about how HAIPA works. Think of it as a library for your agent. If you have a question about how HAIPA works, ask your agent directly. And if your agent gets confused and starts asking you how things work, tell them to get the HAIPA docs first and find out for themselves. **Example prompts** - *"Explain to me how peg orders work."* - *"What parameters does a trailing stop have?"* - *"What's the difference between an algo order and a regular order?"* - *"Don't ask me, dude — check the docs…"*