File size: 23,236 Bytes
80fa9cc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
370010e
 
 
 
 
 
 
 
 
 
 
80fa9cc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
370010e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
80fa9cc
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
# app.py
import os
import re
import asyncio
from datetime import datetime
from zoneinfo import ZoneInfo

import gradio as gr
import pandas as pd

from dotenv import load_dotenv
from agents import Agent, Runner, trace, Tool
from agents.mcp import MCPServerStdio

# Your local helper modules
import hype_accounts_server
from memory_utils import load_memories, save_memory, load_memories_df

load_dotenv(override=True)

# === Time / Locale ===
SGT = ZoneInfo("Asia/Singapore")
def now_sgt():
    return datetime.now(SGT)

# === MCP Server Factories ===
def make_hyperliquid_trader_mcp_servers():
    return [MCPServerStdio(
        {"command": "python3", "args": ["-u", "hype_accounts_server.py"],
         "env": {
             "HYPERLIQUID_API_KEY": os.getenv("HYPERLIQUID_API_KEY"),
             "HYPERLIQUID_PRIVATE_KEY": os.getenv("HYPERLIQUID_PRIVATE_KEY"),
             "HYPERLIQUID_ACCOUNT_ADDRESS": os.getenv("HYPERLIQUID_ACCOUNT_ADDRESS"),
         }},
        client_session_timeout_seconds=30
    )]

def make_crypto_news_mcp_servers():
    # Uses your scraper-based news MCP to avoid API plan limits
    return [MCPServerStdio(
        {"command": "python3", "args": ["-u", "crypto_news_scraper_server.py"]},
        client_session_timeout_seconds=30
    )]

def make_technical_analyst_mcp_servers():
    return [MCPServerStdio(
        {"command": "python3", "args": ["-u", "hl_indicators_server.py"]},
        client_session_timeout_seconds=30
    )]

# === Utils for MCP lifecycle ===
async def connect_all(servers):
    for s in servers:
        await s.connect()

async def close_all(servers):
    for s in servers:
        try:
            await s.close()
        except Exception:
            pass

# === Agent Builders ===
async def build_news_tool(news_servers) -> Tool:
    instructions = (
        "You are a cryptocurrency researcher. You can search and summarise the most relevant, "
        "recent crypto news. If the user asks about a specific coin (e.g., HYPE, BTC, ETH, XRP), "
        "focus on that. Otherwise, highlight notable events and potential long/short opportunities. "
        f"Current datetime (SGT): {now_sgt():%Y-%m-%d %H:%M:%S}."
    )
    agent = Agent(
        name="Crypto news researcher",
        instructions=instructions,
        model="gpt-4.1-mini",
        mcp_servers=news_servers,
    )
    return agent.as_tool(
        tool_name="crypto_news_researcher",
        tool_description="Research crypto news and opportunities for a coin or broad scan."
    )

async def build_ta_tool(ta_servers) -> Tool:
    instructions = (
        "You are a cryptocurrency perpetuals technical trading researcher.\n"
        "Default interval: 1h; default lookback: 36.\n"
        "Indicators: EMA(20,200), MACD(12,26,9), StochRSI(14,14,3,3), ADL, Volume.\n"
        "Given a coin/interval/lookback, compute indicator state, infer trend, and propose entries, "
        "exits, and stop-loss/take-profit with reasoning.\n"
        f"Current datetime (SGT): {now_sgt():%Y-%m-%d %H:%M:%S}."
    )
    agent = Agent(
        name="Crypto technical researcher",
        instructions=instructions,
        model="gpt-4.1-mini",
        mcp_servers=ta_servers,
    )
    return agent.as_tool(
        tool_name="crypto_technical_researcher",
        tool_description="Run TA (EMA, MACD, StochRSI, ADL, Volume)."
    )

async def build_trader(hyper_servers, tools: list[Tool]) -> Agent:
    # Pull short memory + balances so the agent can context-switch well
    past_memories = load_memories(5)
    memory_text = "\n".join(past_memories) if past_memories else "No prior memories."
    try:
        account_details = await hype_accounts_server.get_account_details()
    except Exception as e:
        account_details = f"(Could not fetch account details: {e})"

    instructions = f"""
You are a cryptocurrency perpetuals trader that can:
- Query account balances/positions (via MCP servers on Hyperliquid).
- Do market/news research and TA using attached tools.
- Place long/short orders when the setup has clear edge. Transaction cost: 0.04%.
- If signals are unclear, do NOT trade.

Recent notes:
{memory_text}

Account state:
{account_details}

General rules:
- Prefer confluence: trend + momentum + volume/ADL agreement.
- Always suggest stop-loss and take-profit levels.
- Keep risk per trade modest. Avoid overtrading.
"""
    trader = Agent(
        name="crypto_trader",
        instructions=instructions,
        tools=tools,
        mcp_servers=hyper_servers,  # these expose trading actions
        model="gpt-4.1-mini",
    )
    return trader

# === Intent Routing ===
COMMAND_HELP = """\
You can ask in natural language, e.g.:
β€’ "Balance" / "portfolio" β€” show Hyperliquid balances/positions
β€’ "News on BTC and ETH" β€” market research
β€’ "TA HYPE 1h lookback 48" β€” technical analysis
β€’ "Long HYPE 500 at market, SL 2% TP 4%" β€” execute trade
β€’ "Short BTC 0.01 at 68000, SL 69000 TP 66000" β€” limit order example
β€’ "Summarize opportunities today" β€” broad scan (news + TA)
"""

RE_TA = re.compile(r"\bTA\s+([A-Za-z0-9_\-]+)(?:\s+(\d+[mhHdD]))?(?:\s+lookback\s+(\d+))?", re.IGNORECASE)
RE_LONG = re.compile(r"\bLONG\s+([A-Za-z0-9_\-]+)\s+([\d.]+)(?:\s+at\s+(market|mkt|[\d.]+))?(?:.*?\bSL\s+([\d.%]+))?(?:.*?\bTP\s+([\d.%]+))?", re.IGNORECASE)
RE_SHORT = re.compile(r"\bSHORT\s+([A-Za-z0-9_\-]+)\s+([\d.]+)(?:\s+at\s+(market|mkt|[\d.]+))?(?:.*?\bSL\s+([\d.%]+))?(?:.*?\bTP\s+([\d.%]+))?", re.IGNORECASE)
RE_CLOSE = re.compile(r"\b(close|exit|flatten)\s+(all|[A-Za-z0-9_\-]+)(?:\s+(\d+)%|\s+([\d.]+))?", re.IGNORECASE)

def _close_desc(coin_or_all: str, pct: str | None, qty: str | None) -> str:
    coin_or_all = coin_or_all.upper()
    if coin_or_all == "ALL":
        return "Close ALL open positions at market"
    if pct:
        return f"Close {pct}% of {coin_or_all} position at market"
    if qty:
        return f"Reduce {coin_or_all} position by {qty} units at market"
    return f"Close {coin_or_all} position at market"

def pct_or_price(s):
    if not s:
        return None
    s = s.strip().lower()
    if s.endswith("%"):
        try:
            return {"type": "percent", "value": float(s[:-1])}
        except:
            return None
    try:
        return {"type": "price", "value": float(s)}
    except:
        return None

# === Core Chatbot Handler ===
async def handle_message(message: str, history: list[tuple[str, str]]):
    """
    Routes user intent to: balance, news, TA, or trade execution.
    Returns markdown text.
    """
    text = (message or "").strip()
    ts = now_sgt().strftime("%Y-%m-%d %H:%M:%S %Z")

    # Quick help
    if text.lower() in {"help", "/help", "commands"}:
        return f"### Commands\n{COMMAND_HELP}"

    # 1) Balance / portfolio
    if re.search(r"\b(balance|portfolio|positions?)\b", text, re.IGNORECASE):
        try:
            acct = await hype_accounts_server.get_account_details()
            save_memory(f"[{now_sgt():%Y-%m-%d %H:%M:%S %Z}] User checked balance.")
            return format_account_for_chat(acct)
        except Exception as e:
            return f"❌ Error fetching account details: `{e}`"

    # 2) TA intent
    m = RE_TA.search(text)
    if m:
        coin = m.group(1).upper()
        interval = (m.group(2) or "1h").lower()
        lookback = int(m.group(3) or 36)

        news_servers = []  # not needed here
        ta_servers = []
        try:
            ta_servers = make_technical_analyst_mcp_servers()
            await connect_all(ta_servers)

            ta_tool = await build_ta_tool(ta_servers)

            # Build a "TA-only" agent so we don't touch trading MPC here
            researcher = Agent(
                name="crypto_ta_agent",
                instructions=f"Focus on TA for {coin} at interval {interval}, lookback {lookback}. Output indicator values and strategy.",
                tools=[ta_tool],
                model="gpt-4.1-mini",
            )

            prompt = f"Run TA for {coin} on {interval}, lookback {lookback}. Return indicators and actionable plan."
            with trace("crypto_ta"):
                result = await Runner.run(researcher, prompt, max_turns=12)

            save_memory(f"[{ts}] TA {coin} {interval} lookback {lookback}")
            return f"### πŸ”¬ TA β€” {coin} ({interval}, lookback {lookback})\n\n{result.final_output}"
        except Exception as e:
            return f"❌ TA error: `{e}`"
        finally:
            await close_all(ta_servers)

    # 3) Trade intent (LONG / SHORT)
    mm = RE_LONG.search(text) or RE_SHORT.search(text)
    if mm:
        is_long = bool(RE_LONG.search(text))
        side = "LONG" if is_long else "SHORT"
        coin = mm.group(1).upper()
        qty = float(mm.group(2))
        at = mm.group(3)  # "market"/"mkt" or price
        sl_raw = mm.group(4)
        tp_raw = mm.group(5)
        sl = pct_or_price(sl_raw)
        tp = pct_or_price(tp_raw)

        price_desc = "market" if (at is None or str(at).lower() in {"market", "mkt"}) else at
        order_desc = f"{side} {coin} {qty} at {price_desc}"
        if sl: order_desc += f", SL {sl_raw}"
        if tp: order_desc += f", TP {tp_raw}"

        hyper_servers = []
        news_servers = []
        ta_servers = []
        try:
            # Tools available to the *trader*: news + TA
            news_servers = make_crypto_news_mcp_servers()
            ta_servers = make_technical_analyst_mcp_servers()
            hyper_servers = make_hyperliquid_trader_mcp_servers()

            await asyncio.gather(
                connect_all(news_servers),
                connect_all(ta_servers),
                connect_all(hyper_servers),
            )

            news_tool = await build_news_tool(news_servers)
            ta_tool = await build_ta_tool(ta_servers)

            trader = await build_trader(hyper_servers, [news_tool, ta_tool])

            # Natural-language trade instruction to the trader agent.
            trade_prompt = f"""
            User requested: {order_desc}.
            If safe and reasonable given risk rules, place the order via Hyperliquid MCP.
            - If price specified (numeric), treat as limit; otherwise market.
            - Always include stop-loss and take-profit (convert % to prices).
            - Confirm the exact order(s) you placed and rationale in the output.
            """
            with trace("trade_execution"):
                result = await Runner.run(trader, trade_prompt, max_turns=20)

            save_memory(f"[{ts}] Executed: {order_desc}")
            return f"### 🧾 Execution β€” {order_desc}\n\n{result.final_output}"
        except Exception as e:
            return f"❌ Trade execution error: `{e}`"
        finally:
            await asyncio.gather(
                close_all(news_servers),
                close_all(ta_servers),
                close_all(hyper_servers),
            )

    # 4) News intent (e.g., "news on BTC", "what's happening to HYPE")
    if re.search(r"\b(news|headline|what's happening|what is happening|happening)\b", text, re.IGNORECASE):
        # Try to pick coins mentioned
        coins = re.findall(r"\b([A-Z]{2,6})\b", text.upper())
        coins = [c for c in coins if c not in {"NEWS", "HELP"}]
        topic = ", ".join(coins) if coins else "broad market"
        news_servers = []
        try:
            news_servers = make_crypto_news_mcp_servers()
            await connect_all(news_servers)
            news_tool = await build_news_tool(news_servers)

            researcher = Agent(
                name="crypto_news_agent",
                instructions=f"Focus news on: {topic}. Be concise and actionable.",
                tools=[news_tool],
                model="gpt-4.1-mini",
            )
            prompt = f"Summarize the most relevant crypto news for {topic}. Include potential trade angles."
            with trace("crypto_news"):
                result = await Runner.run(researcher, prompt, max_turns=12)

            save_memory(f"[{ts}] News requested: {topic}")
            return f"### πŸ—žοΈ News β€” {topic}\n\n{result.final_output}"
        except Exception as e:
            return f"❌ News error: `{e}`"
        finally:
            await close_all(news_servers)

    # 5) Summary scan (news + TA picks)
    if re.search(r"\b(opportunit|ideas|setup|summary|today)\b", text, re.IGNORECASE):
        hyper_servers = []
        news_servers = []
        ta_servers = []
        try:
            news_servers = make_crypto_news_mcp_servers()
            ta_servers = make_technical_analyst_mcp_servers()
            hyper_servers = make_hyperliquid_trader_mcp_servers()
            await asyncio.gather(
                connect_all(news_servers),
                connect_all(ta_servers),
                connect_all(hyper_servers),
            )
            news_tool = await build_news_tool(news_servers)
            ta_tool = await build_ta_tool(ta_servers)
            trader = await build_trader(hyper_servers, [news_tool, ta_tool])

            prompt = (
                "Step 1: Broad news scan for major catalysts.\n"
                "Step 2: Pick 3–5 coins with potential edges; run compact TA summary (1h, lookback 36).\n"
                "Step 3: Recommend 1–2 best setups with entry, SL, TP and rationale. Do NOT place orders."
            )
            with trace("daily_opportunities"):
                result = await Runner.run(trader, prompt, max_turns=24)

            save_memory(f"[{ts}] Opportunity summary requested.")
            return f"### πŸ“Œ Opportunities β€” {ts}\n\n{result.final_output}"
        except Exception as e:
            return f"❌ Summary error: `{e}`"
        finally:
            await asyncio.gather(
                close_all(news_servers),
                close_all(ta_servers),
                close_all(hyper_servers),
            )

    # Fallback: clarify + brief help
    return (
        "I can help with balance, news, TA, and trade execution.\n\n"
        + COMMAND_HELP
    )

    mclose = RE_CLOSE.search(text)
    if mclose:
        # groups: verb, coin_or_all, pct (digits%), qty (number)
        coin_or_all = mclose.group(2).strip()
        pct = mclose.group(3)  # e.g., "50" meaning 50%
        qty = mclose.group(4)  # absolute size to reduce
        desc = _close_desc(coin_or_all, pct, qty)
    
        hyper_servers = []
        news_servers = []
        ta_servers = []
        try:
            # Tools for trader context (optional but helpful)
            news_servers = make_crypto_news_mcp_servers()
            ta_servers   = make_technical_analyst_mcp_servers()
            hyper_servers = make_hyperliquid_trader_mcp_servers()
    
            await asyncio.gather(
                connect_all(news_servers),
                connect_all(ta_servers),
                connect_all(hyper_servers),
            )
    
            news_tool = await build_news_tool(news_servers)
            ta_tool   = await build_ta_tool(ta_servers)
            trader    = await build_trader(hyper_servers, [news_tool, ta_tool])
    
            # Natural-language prompt to place the close orders via Hyperliquid MCP
            # (The trader agent already has the rules + account context)
            trade_prompt = f"""
            User request: {desc}.
            Instructions:
            - If 'ALL', close every open position at market.
            - If a coin is specified:
                - If a percent is provided, close that % of the CURRENT open position size.
                - If a qty is provided, reduce by that absolute base-asset amount.
                - If neither provided, fully close that coin position.
            - Include SL/TP cleanup if needed (cancel/replace any attached orders).
            - If the coin has no open position, report that clearly.
            - Return a concise execution summary listing each order (coin, side, size, order type, price if applicable) and rationale.
            """
            with trace("close_positions"):
                result = await Runner.run(trader, trade_prompt, max_turns=20)
    
            save_memory(f"[{now_sgt():%Y-%m-%d %H:%M}] Close: {desc}")
            return f"### 🧹 Close β€” {desc}\n\n{result.final_output}"
        except Exception as e:
            return f"❌ Close error: `{e}`"
        finally:
            await asyncio.gather(
                close_all(news_servers),
                close_all(ta_servers),
                close_all(hyper_servers),
            )

# ---------- Pretty printing for account/positions ----------

from math import isnan

def _fnum(x, decimals=2):
    try:
        v = float(x)
        return f"{v:,.{decimals}f}"
    except Exception:
        return str(x)

def _fpct(x, decimals=2):
    try:
        v = float(x) * 100  # input is ROE like 0.0036 -> 0.36%
        sign = "🟒" if v > 0 else ("πŸ”΄" if v < 0 else "βšͺ️")
        return f"{sign} {v:.{decimals}f}%"
    except Exception:
        return "β€”"

def _pnl(x, decimals=2):
    try:
        v = float(x)
        sign = "🟒" if v > 0 else ("πŸ”΄" if v < 0 else "βšͺ️")
        return f"{sign} ${abs(v):,.{decimals}f}"
    except Exception:
        return "β€”"

def _side_and_abs_size(szi):
    try:
        v = float(szi)
        side = "LONG" if v > 0 else ("SHORT" if v < 0 else "FLAT")
        return side, abs(v)
    except Exception:
        return "β€”", szi

def format_account_for_chat(acct: dict) -> str:
    """
    Converts the get_account_details() dict into a nice Markdown summary.
    """
    if not isinstance(acct, dict):
        return f"```\n{acct}\n```"

    holdings = acct.get("holdings", []) or []
    cash = acct.get("cash_balance", "0")
    realized_pnl = acct.get("profit_and_loss", None)

    # Totals
    total_pos_value = 0.0
    total_margin_used = 0.0
    total_upnl = 0.0

    rows_md = []
    for h in holdings:
        pos = h.get("position", {})
        coin = pos.get("coin", "β€”")
        szi = pos.get("szi", 0)
        side, abs_size = _side_and_abs_size(szi)
        entry = pos.get("entryPx", "β€”")
        pval = pos.get("positionValue", 0)
        u = pos.get("unrealizedPnl", 0)
        roe = pos.get("returnOnEquity", 0)
        lev = pos.get("leverage", {})
        lev_str = f"{lev.get('type','β€”')}Γ—{lev.get('value','β€”')}"
        liq = pos.get("liquidationPx", None)
        m_used = pos.get("marginUsed", 0)
        fund = pos.get("cumFunding", {}).get("sinceOpen", None)

        # Totals
        try: total_pos_value += float(pval)
        except: pass
        try: total_margin_used += float(m_used)
        except: pass
        try: total_upnl += float(u)
        except: pass

        rows_md.append(
            f"| {coin} | {side} | {_fnum(abs_size, 6)} | ${_fnum(entry, 2)} | ${_fnum(pval, 2)} | {_pnl(u, 2)} | {_fpct(roe, 2)} | {lev_str} | {('β€”' if liq in (None, 'None') else '$'+_fnum(liq, 2))} | ${_fnum(m_used, 2)} | {('β€”' if fund in (None, 'None') else _fnum(fund, 6))} |"
        )

    header = (
        "### πŸ“Š Account / Positions\n"
        f"- **Cash balance:** ${_fnum(cash, 2)}\n"
        f"- **Total pos. value:** ${_fnum(total_pos_value, 2)}\n"
        f"- **Unrealized PnL:** {_pnl(total_upnl, 2)}\n"
        f"- **Margin used (total):** ${_fnum(total_margin_used, 2)}\n"
    )
    if realized_pnl is not None:
        header += f"- **Realized PnL (session/period):** {_pnl(realized_pnl, 2)}\n"

    table_head = (
        "\n| Coin | Side | Size | Entry Px | Pos. Value | uPnL | ROE | Leverage | Liq Px | Margin Used | Funding (since open) |\n"
        "|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|\n"
    )
    table_body = "\n".join(rows_md) if rows_md else "_No open positions_"

    return header + table_head + table_body


# === Gradio UI ===
with gr.Blocks(fill_height=True) as demo:
    gr.Markdown("# πŸ€– Crypto Trading Copilot")
    gr.Markdown(
        f"Local time: **{now_sgt():%Y-%m-%d %H:%M:%S %Z}**  \n"
        "[OpenAI Traces](https://platform.openai.com/logs?api=traces) Β· "
        "[Hyperliquid](https://app.hyperliquid.xyz/trade)"
    )

    with gr.Row():
        quick1 = gr.Button("πŸ“Š Balance")
        quick2 = gr.Button("πŸ—žοΈ News: BTC, ETH")
        quick3 = gr.Button("πŸ”¬ TA: HYPE 1h")
        quick4 = gr.Button("🧾 Long HYPE 500 @ market (SL 2% TP 4%)")

    chatbot = gr.Chatbot(height=480, type="messages", show_copy_button=True)
    user_in = gr.Textbox(placeholder="Try: TA HYPE 1h lookback 48  β€’  News on BTC  β€’  Long HYPE 500 at market, SL 2% TP 4%", scale=1)
    send_btn = gr.Button("Send", variant="primary")

    with gr.Accordion("Memory (last 10)", open=False):
        mem_table = gr.Dataframe(value=load_memories_df(10), interactive=False, wrap=True, show_label=False)

    async def _respond(user_msg, chat_state):
        bot_md = await handle_message(user_msg, chat_state or [])
        # Log short memory line
        save_memory(f"[{now_sgt():%Y-%m-%d %H:%M}] {user_msg[:80]}")
        # Update display memory table
        latest_mem = load_memories_df(10)
        return chat_state + [{"role":"user","content":user_msg},{"role":"assistant","content":bot_md}], "", latest_mem

    send_btn.click(_respond, inputs=[user_in, chatbot], outputs=[chatbot, user_in, mem_table])
    user_in.submit(_respond, inputs=[user_in, chatbot], outputs=[chatbot, user_in, mem_table])

    # Quick actions
    async def _qa_balance(chat_state):
        msg = "balance"
        bot_md = await handle_message(msg, chat_state or [])
        save_memory(f"[{now_sgt():%Y-%m-%d %H:%M}] Quick: balance")
        latest_mem = load_memories_df(10)
        return chat_state + [{"role":"user","content":msg},{"role":"assistant","content":bot_md}], latest_mem

    async def _qa_news(chat_state):
        msg = "news on BTC and ETH"
        bot_md = await handle_message(msg, chat_state or [])
        save_memory(f"[{now_sgt():%Y-%m-%d %H:%M}] Quick: news BTC ETH")
        latest_mem = load_memories_df(10)
        return chat_state + [{"role":"user","content":msg},{"role":"assistant","content":bot_md}], latest_mem

    async def _qa_ta(chat_state):
        msg = "TA HYPE 1h lookback 48"
        bot_md = await handle_message(msg, chat_state or [])
        save_memory(f"[{now_sgt():%Y-%m-%d %H:%M}] Quick: TA HYPE")
        latest_mem = load_memories_df(10)
        return chat_state + [{"role":"user","content":msg},{"role":"assistant","content":bot_md}], latest_mem

    async def _qa_long(chat_state):
        msg = "Long HYPE 500 at market, SL 2% TP 4%"
        bot_md = await handle_message(msg, chat_state or [])
        save_memory(f"[{now_sgt():%Y-%m-%d %H:%M}] Quick: long HYPE")
        latest_mem = load_memories_df(10)
        return chat_state + [{"role":"user","content":msg},{"role":"assistant","content":bot_md}], latest_mem

    quick1.click(_qa_balance, inputs=[chatbot], outputs=[chatbot, mem_table])
    quick2.click(_qa_news, inputs=[chatbot], outputs=[chatbot, mem_table])
    quick3.click(_qa_ta, inputs=[chatbot], outputs=[chatbot, mem_table])
    quick4.click(_qa_long, inputs=[chatbot], outputs=[chatbot, mem_table])

if __name__ == "__main__":
    # No deprecated args; queue() OK without concurrency_count
    demo.queue().launch()