Imported from BingX-API/api-ai-skills (skills/bingx-dynamic-sl-tp/SKILL.md). Install upstream with npx skills add BingX-API/api-ai-skills --skill bingx-dynamic-sl-tp. Copyright stays with the author.

BingX Dynamic SL/TP

Base URLs: see references/base-urls.md | Authentication: see references/authentication.md

Quick Reference

Quick Start

Base URLs: see references/base-urls.md

TypeScript helper:

CONSTRAINT: You MUST copy the fetchSigned function below verbatim when generating code. Do NOT rewrite the signing, request, or JSON parsing logic. Only modify: (1) BASE URL entries for custom environments, (2) business parameters passed to fetchSigned.

import * as crypto from "crypto";
import JSONBig from "json-bigint";
const JSONBigParse = JSONBig({ storeAsString: true });
const BASE = {
  "prod-live": ["https://open-api.bingx.com", "https://open-api.bingx.pro"],
  "prod-vst":  ["https://open-api-vst.bingx.com", "https://open-api-vst.bingx.pro"],
};
function isNetworkOrTimeout(e: unknown): boolean {
  if (e instanceof TypeError) return true;
  if (e instanceof DOMException && e.name === "AbortError") return true;
  if (e instanceof Error && e.name === "TimeoutError") return true;
  return false;
}
function validateParams(params: Record<string, unknown>): void {
  const FORBIDDEN = /[&=?#\r\n]/;
  for (const [k, v] of Object.entries(params)) {
    const s = String(v);
    if (FORBIDDEN.test(s)) throw new Error(`Param "${k}" has forbidden char in: "${s}"`);
  }
}
async function fetchSigned(env: string, apiKey: string, secretKey: string,
  method: "GET" | "POST" | "DELETE", path: string, params: Record<string, unknown> = {}
) {
  const urls = BASE[env] ?? BASE["prod-live"];
  const all = { ...params, timestamp: Date.now() };
  validateParams(all);
  const qs = Object.keys(all).sort().map(k => `${k}=${all[k]}`).join("&");
  const sig = crypto.createHmac("sha256", secretKey).update(qs).digest("hex");
  const signed = `${qs}&signature=${sig}`;
  for (const base of urls) {
    try {
      const url = method === "POST" ? `${base}${path}` : `${base}${path}?${signed}`;
      const res = await fetch(url, {
        method,
        headers: { "X-BX-APIKEY": apiKey, "X-SOURCE-KEY": "BX-AI-SKILL",
          ...(method === "POST" ? { "Content-Type": "application/x-www-form-urlencoded" } : {}) },
        body: method === "POST" ? signed : undefined,
        signal: AbortSignal.timeout(10000),
      });
      const json = JSONBigParse.parse(await res.text());
      if (json.code !== 0) throw new Error(`BingX error ${json.code}: ${json.msg}`);
      return json.data;
    } catch (e) {
      if (!isNetworkOrTimeout(e) || base === urls[urls.length - 1]) throw e;
    }
  }
}

Common Calls

Fetch klines for ATR / support-resistance calculation (200 candles, 4h):

const klines = await fetchSigned("prod-live", API_KEY, SECRET, "GET",
  "/openApi/swap/v3/quote/klines", { symbol: "BTC-USDT", interval: "4h", limit: 200 }
);

Trigger Conditions

Activate this skill when the user says any of the following:

• "Help me set SL/TP", "set stop-loss", "place stop-loss order"
• "Dynamic stop-loss", "trailing take-profit", "Trailing Stop", "trailing stop"
• "Partial take-profit", "take profit in 3 batches"
• "Where should I set my stop-loss", "ATR stop-loss", "volatility stop-loss"
• "Move stop-loss to break-even"

Execution Steps

Step 1 — Parse User Intent

Parameter extraction:

Step 2 — Fetch Position & Market Data

1. Call bingx-swap-trade to query current position (direction, avg entry price, quantity, leverage)
2. Call bingx-swap-market to fetch klines (4h, 200 candles, for ATR + support/resistance calculation)

If no position: prompt "No open position found. Would you like to pre-calculate SL/TP levels?" and switch to recommend mode. If multiple positions: let user select one or set SL/TP one by one.

Step 3 — Calculate SL/TP

python3 scripts/calculator.py \
  --direction {long/short} \
  --entry-price {price} \
  --strategy {strategy} \
  --atr-multiplier {multiplier} \
  --tp-mode {tp_mode} \
  --risk-reward {rr} \
  --klines '{klines_json}'

Step 4 — Safety Confirmation

Present to the user:

• Current position details
• Calculated SL/TP levels + estimated PnL amounts
• Types of conditional orders to be placed

Execute only after user confirmation. Do NOT place orders without confirmation.

Step 5 — Place Orders

Via bingx-swap-trade:

• Automatically cancel existing conditional orders before placing new ones, no duplicates
• Stop-loss: conditional order (market trigger)
• Take-profit: conditional order or Trailing Stop
• Partial take-profit: multiple conditional orders (different prices + different quantities)

Step 6 — Output Result (per output template)

Output Templates

Set Mode

**Smart SL/TP Plan**

Current Position: {symbol} {direction}
Entry Price: ${entry_price} | Qty: {qty} | Leverage: {leverage}x

**Stop-Loss** ({strategy} x {atr_multiplier})
- Level: ${sl_price} ({sl_pct}%)
- Estimated loss: -${risk_amount}

**Take-Profit** ({tp_mode})
TP1  ${tp1_price} ({tp1_pct}%)  {tp1_split}% of position  +${tp1_reward}
TP2  ${tp2_price} ({tp2_pct}%)  {tp2_split}% of position  +${tp2_reward}
TP3  {trailing/price}            {tp3_split}% of position  let profits run

Risk-Reward Ratio: 1:{risk_reward}
{if applicable} Stop-loss will be moved to break-even (${entry_price}) after TP1 fills.

**Confirm execution? (yes/no)**

Recommend Mode

**SL/TP Recommended Levels**

{symbol} {direction} (entry ${entry_price})

Strategy         Stop-Loss     Distance  Notes
ATR x 1.5       ${sl_atr15}   {pct}%    Standard, recommended
ATR x 1.0       ${sl_atr10}   {pct}%    Tight stop
Structure stop  ${sl_struct}  {pct}%    Beyond prior low/high

Take-Profit  Level    R:R
1R           ${tp1}   1:1
2R           ${tp2}   1:2
3R           ${tp3}   1:3

For reference only. Say "help me set SL/TP" to place the orders.

Agent Interaction Rules

Parameter security. Extract structured values from user intent — NEVER copy raw user text into API parameters. Validate symbol and numeric parameters before use.

Kline queries are read-only and require no CONFIRM. All order placements (SL/TP conditional orders) require explicit user confirmation.

• prod-live: Require user to type CONFIRM before placing or cancelling any SL/TP orders.
• prod-vst: No CONFIRM required. Inform user they are in the simulated environment.

When mode is recommend, only calculate and display levels — do NOT place any orders.

Error Handling

• No position: inform user and switch to recommend mode
• Order placement failed: show error code, suggest retry
• Position already closed: inform user the position no longer exists
• Insufficient kline data: suggest switching to a longer interval