Bloxwap's signals are available to AI agents. Through the Bloxwap MCP server, an agent can read the same Trend marks and Swing setups you see in the app, as structured data it can filter, compare, and explain.

This guide covers the two signal tools, what each field means, and how to have an agent turn a signal into a trade plan that respects limits you set.

The agent reads signals. It does not trade on them. The server's order tools are not available at this time, so you place every trade yourself. See When a signal fires for the ways to act.

If you haven't connected an agent yet, the setup guide takes a minute. Both signal tools are public, so there's nothing to sign in to.

The two signal tools

ToolWhat it returns
get_trend_signalsThe current Trend state for each market with a recent reading.
get_swing_signalsEvery active Swing setup, with direction, leverage, and three price levels.

For how the signals themselves work, read Bloxwap Signals: how Swing and Trend work. In short, Trend describes a move that has already happened, and Swing lays out a complete trade setup.

What a Trend signal contains

get_trend_signals takes no arguments. Each entry looks like this:

{
  "coin": "BTC",
  "state": "UP",
  "timestamp": 1794268200000
}
FieldMeaning
coinThe market symbol.
stateUP, DOWN, WATCH, or NO_SIGNAL.
timestampWhen the state was observed, in milliseconds since 1 January 1970 (UTC).
confidenceAn optional score from 0 to 1. It is not always present.

UP and DOWN mark a large, efficient 15-minute move on expanding volatility. WATCH means volatility is expanding with no clear direction yet. NO_SIGNAL means none of those conditions hold.

The tool returns only readings from the last five minutes. A market that's missing from the list has no fresh reading, which is different from a NO_SIGNAL state.

What a Swing signal contains

get_swing_signals accepts three optional filters: coin, direction (LONG or SHORT), and tradableOnly. Each setup looks like this:

{
  "coin": "BTC",
  "signalId": "example-signal-id",
  "family": "example-family",
  "direction": "LONG",
  "entryPrice": 60000,
  "stopLoss": 59400,
  "takeProfit": 61200,
  "leverage": 10,
  "confidence": 0.8,
  "edgeScore": null,
  "timestamp": 1794268200000,
  "expiresAt": 1794269100000,
  "tradable": true,
  "entryPriceMin": 59950,
  "entryPriceMax": 60050
}

These values are an illustration, not a live signal.

FieldMeaning
directionLONG benefits from a rise in price; SHORT benefits from a fall.
entryPriceThe reference price for entering the setup.
stopLossThe intended exit if price moves against the position.
takeProfitThe intended exit if price reaches the target.
leverageThe leverage the setup suggests.
confidenceA score from 0 to 1, or null. It is not a probability of profit.
expiresAtWhen the setup stops being valid, in milliseconds since 1 January 1970 (UTC).
tradablefalse marks a test signal. Pass tradableOnly: true to leave these out.
entryPriceMin, entryPriceMaxThe price range in which the entry still applies, when the setup has one.
signalId, familyIdentifiers for the setup and the strategy that produced it.
edgeScoreAn optional internal ranking score, or null.

In the app, Swing is a Genesis feature and confidence appears as one to five bars.

Example: watch Trend on BTC

Here's a prompt that puts the Trend tool to work:

Check Trend on BTC every few minutes. When it flips to UP, tell me the current price and the most leverage I could use.

When the agent checks, it calls get_trend_signals and finds the BTC entry. If the state is UP, it calls get_market_prices for the price and list_markets for the leverage cap, then reports back.

Two things matter here:

  • The agent has to be running. A chat agent checks only when you ask. To check on a schedule, use an agent or client that supports recurring tasks. For alerts that reach you on their own, turn on signal notifications in the app with the notifications guide.
  • UP describes the past. A Trend mark says a strong move just happened. It doesn't predict that the move will continue, and a pullback can follow. Treat the alert as a reason to look, not a reason to buy.

Turn a Swing setup into a trade plan

A Swing setup already has an entry, a stop, and a target. Your agent's job is to check the setup against the live market and against your limits:

Get the tradable Swing setups for BTC. For each one, tell me whether it has expired, whether the current price is still inside the entry range, the distance to the stop and to the target as a percentage, and what a round trip would cost with $50 of margin.

For the illustrated setup above, the agent would work through it like this:

  1. Is it still valid? Compare expiresAt with the current time. An expired setup is history.
  2. Is the price still in range? Compare the live price with entryPriceMin and entryPriceMax. If price has already run past the range, the setup's levels no longer describe the trade you'd get.
  3. Where are the exits? The stop at $59,400 is 1% below the $60,000 entry. The target at $61,200 is 2% above it.
  4. What does the leverage do to that? At 10× leverage, a 1% move against you is about 10% of your margin, before fees. With $50 of margin, that's roughly $5 lost at the stop and $10 gained at the target.
  5. What does it cost? quote_fees gives the round-trip fees for that margin and leverage.

For a SHORT setup the stop sits above the entry and the target sits below.

Stop-loss and take-profit prices are triggers, not guaranteed fill prices. In a fast market the exit can fill at a worse price than the level you set.

Give your agent risk limits

An agent follows the rules you write down. Put your limits in the prompt, or in the agent's standing instructions, so every plan gets checked against them:

Only show me setups that fit these rules: no more than $50 of margin per trade, leverage at or below 5× even if the signal suggests more, a stop no further than 1.5% from entry, and nothing that expires in under ten minutes. Skip any market where I already have a position.

A few limits worth setting:

  • Margin per trade: the most you're willing to put up on one idea.
  • Your own leverage cap: a signal's suggested leverage and the market's maximum are both ceilings. You can always use less.
  • Stop distance: how far price can move against you before you're out.
  • One position per market: with a signed-in connection, the agent can call get_positions to see what you already hold.

These limits shape what the agent tells you. They don't bind your account, because the agent can't place or block a trade. You're the last check.

When a signal fires

The agent can find and vet a setup. Acting on it is up to you, and there are three ways to do that today:

  • Place the trade in the app. Open the market in Bloxwap, set your amount and leverage, and add the stop and target from the plan.
  • Use the Swing push notification. On supported devices, a Swing notification offers an Open action that can submit a live position with the signal's direction, leverage, and exits, using your selected trade amount. It's opt-in: you enable Swing and its notifications first. The Signals guide explains exactly what Open does.
  • Build a bot. Developers who want fully automated execution can trade through the Hyperliquid SDK with a dedicated agent wallet. Read agent permissions first: an approved agent wallet can trade without per-order limits, so the risk rules have to live in your own code.

Automatic execution through the MCP server is not available at this time. We'll turn the order tools on only when an agent can sign with a narrowly scoped key of its own.

Keep the risk in view

No signal guarantees a profitable trade, and an agent reading a signal doesn't change that. Leverage magnifies losses as well as gains. Fees, funding, and liquidation risk still apply, and a stop-loss can't guarantee a maximum loss.

Agents also make mistakes. They can misread a field, do the arithmetic wrong, or describe a setup that has since expired. Check the price, the levels, and your position in the app before you act.

If you're new to these ideas, start with what perpetual futures are, review the fees, and learn how to set a stop-loss.