File size: 70,251 Bytes
72282ab 780c7ab 72282ab | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 | #!/usr/bin/env python3
"""
Enhanced MCP Client with LLM-based task decomposition, intelligent agent routing, and real MCP protocol.
This client uses AI for smart query analysis and agent coordination instead of hard-coded rules.
"""
import gradio as gr
import asyncio
import json
import os
import sys
import argparse
from typing import Dict, List, Any, Optional
from dataclasses import dataclass
from enum import Enum
import uuid
from pathlib import Path
import requests
from smolagents import MCPClient, LiteLLMModel
ANTHROPIC_API_KEY = os.environ.get('ANTHROPIC_API_KEY')
def setup_environment():
"""Set up environment variables and configuration."""
global ANTHROPIC_API_KEY
# Validate API keys
print("\n🔑 API Configuration:")
print(f"Anthropic API Key: {'✓ Configured' if ANTHROPIC_API_KEY else '✗ Missing'}")
if not ANTHROPIC_API_KEY:
print("⚠️ Warning: ANTHROPIC_API_KEY not found in environment")
print("💡 Set environment variable: ANTHROPIC_API_KEY=your_anthropic_key")
class TaskType(Enum):
"""Types of tasks that can be decomposed."""
SENTIMENT_ANALYSIS = "sentiment_analysis"
LOCATION_SEARCH = "place_search"
RESTAURANT_SEARCH = "restaurant_search"
HIKING_SEARCH = "hiking_search"
WEB_SEARCH = "web_search"
COMPLEX_QUERY = "complex_query"
@dataclass
class SubTask:
"""Represents a sub-atomic task."""
id: str
task_type: TaskType
description: str
parameters: Dict[str, Any]
agent_id: str
confidence: float = 0.5
status: str = "pending"
result: Optional[Dict[str, Any]] = None
@dataclass
class Agent:
"""Represents a dedicated agent for handling specific tools."""
id: str
name: str
tool_name: str
description: str
capabilities: List[str]
keywords: List[str]
class LLMTaskDecomposer:
"""LLM-powered task decomposer using system prompts for intelligent query analysis."""
def __init__(self, model_name: str = "anthropic"):
"""Initialize with support for multiple LLM providers."""
self.model_name = model_name.lower()
self.model = None
# Initialize the model based on selection
self._initialize_model()
self.agents = self._initialize_agents()
def _initialize_model(self):
"""Initialize the selected model with proper error handling."""
try:
if self.model_name == "anthropic":
if not ANTHROPIC_API_KEY:
print("❌ ANTHROPIC_API_KEY environment variable is required for Anthropic model")
print("💡 Model will fall back to keyword-based decomposition")
self.model = None
return
print(f"🔧 Initializing Anthropic model...")
self.model = LiteLLMModel(
model_id="anthropic/claude-sonnet-4-20250514",
temperature=0.2,
api_key=ANTHROPIC_API_KEY
)
# Test the model with a simple call
try:
test_response = self.model([{"role": "user", "content": "Hello"}])
print(f"✅ Anthropic model initialized and tested successfully")
print(f"🧠 Model response test: {str(test_response)[:50]}...")
except Exception as test_error:
print(f"⚠️ Model initialized but test call failed: {test_error}")
print(f"🔄 Will attempt to use model anyway, with fallback to keywords")
else:
print(f"❌ Unknown model name: {self.model_name}")
self.model = None
except Exception as e:
print(f"❌ Model initialization failed: {e}")
print(f"🔄 Falling back to keyword-based decomposition")
self.model = None
def get_model_info(self) -> Dict[str, str]:
"""Get information about the current model."""
if self.model_name == "anthropic":
return {
"name": "Claude Sonnet 4",
"provider": "Anthropic",
"emoji": "🤖",
"model_id": "anthropic/claude-sonnet-4-20250514",
"status": "initialized" if self.model else "failed"
}
else:
return {
"name": "Unknown Model",
"provider": "Unknown",
"emoji": "❓",
"model_id": "unknown",
"status": "failed"
}
def _initialize_agents(self) -> Dict[str, Agent]:
"""Initialize specialized agents with their capabilities and keywords."""
agents = {
"sentiment_agent": Agent(
id="sentiment_agent",
name="Sentiment Analysis Agent",
tool_name="sentiment_analysis",
description="Analyzes text sentiment, emotions, and opinions",
capabilities=["text_analysis", "emotion_detection", "polarity_scoring", "opinion_mining"],
keywords=["sentiment", "feeling", "opinion", "review", "emotion", "mood", "analyze text", "positive", "negative", "happy", "sad", "angry", "excited"]
),
"location_agent": Agent(
id="location_agent",
name="Location Search Agent",
tool_name="place_search",
description="Finds hotels, accommodations, and places to stay",
capabilities=["place_search", "hotel_finder", "accommodation_search", "lodging_recommendations"],
keywords=["hotel", "hotels", "stay", "accommodation", "lodging", "motel", "resort", "inn", "bed and breakfast", "airbnb", "place to stay"]
),
"restaurant_agent": Agent(
id="restaurant_agent",
name="Restaurant Search Agent",
tool_name="restaurant_search",
description="Discovers restaurants, food places, and dining options",
capabilities=["restaurant_search", "cuisine_finder", "dining_recommendations", "food_discovery"],
keywords=["restaurant", "restaurants", "food", "dining", "eat", "dinner", "lunch", "breakfast", "cafe", "bar", "cuisine", "meal", "dining out"]
),
"hiking_agent": Agent(
id="hiking_agent",
name="Hiking Search Agent",
tool_name="hiking_search",
description="Finds hiking trails, outdoor activities, and nature spots",
capabilities=["trail_finder", "outdoor_activities", "difficulty_assessment", "nature_exploration"],
keywords=["hike", "hiking", "trail", "trails", "trek", "trekking", "outdoor", "mountain", "nature", "walk", "walking", "climbing", "adventure"]
),
"web_agent": Agent(
id="web_agent",
name="Web Search Agent",
tool_name="web_search",
description="Searches web for information, news, weather, finance, and general queries with intelligent ticker detection for financial data",
capabilities=["web_search", "information_retrieval", "real_time_data", "news_search", "weather_data", "financial_data", "ticker_detection"],
keywords=["search", "find", "lookup", "google", "web", "information", "weather", "news", "current", "latest", "what is", "definition", "stock", "price", "market", "finance"]
)
}
return agents
async def decompose_query(self, user_query: str) -> List[SubTask]:
"""
Use LLM to analyze user query and decompose into actionable subtasks.
"""
print(f"🔍 Decomposing query: '{user_query}'")
try:
# Create decomposition prompt
system_prompt = self._create_decomposition_prompt()
# Prepare the user message
user_message = f"""Query to analyze: "{user_query}"
Please analyze this query and respond with a JSON object containing your analysis."""
print(f"🧠 Attempting LLM decomposition with model: {self.model_name}")
# Use Anthropic model (synchronous)
if self.model_name == "anthropic" and self.model is not None:
# Use LiteLLM model directly (synchronous)
try:
print(f"📡 Calling LLM model...")
response = self.model([
{"role": "system", "content": system_prompt},
{"role": "user", "content": user_message}
])
print(f"✅ LLM response received: {str(response)[:200]}...")
print(f"🔍 Response type: {type(response)}")
except Exception as e:
print(f"❌ Model call failed: {e}")
print(f"🔄 Falling back to keyword-based decomposition")
return self._fallback_decomposition(user_query)
else:
print(f"❌ Model not available ({self.model_name}), using fallback")
return self._fallback_decomposition(user_query)
# Parse LLM response
print(f"🔍 Parsing LLM response...")
analysis = self._parse_llm_response(response, user_query)
print(f"📋 Analysis result: {analysis}")
# Convert analysis to subtasks
print(f"🎯 Creating subtasks from analysis...")
subtasks = self._create_subtasks(analysis, user_query)
print(f"✅ Generated {len(subtasks)} subtasks")
if not subtasks:
print("⚠️ No subtasks generated, using fallback")
return self._fallback_decomposition(user_query)
# Debug: print subtask details
for i, subtask in enumerate(subtasks):
print(f" 📝 Subtask {i+1}: {subtask.agent_id} -> {subtask.description}")
return subtasks
except Exception as e:
print(f"❌ Task decomposition failed: {e}")
print(f"🔄 Using fallback decomposition")
return self._fallback_decomposition(user_query)
def _create_decomposition_prompt(self) -> str:
"""Create comprehensive system prompt for task decomposition."""
agent_descriptions = []
for agent_id, agent in self.agents.items():
agent_descriptions.append(f"""
**{agent.name}** ({agent_id}):
- Description: {agent.description}
- Tool: {agent.tool_name}
- Keywords: {', '.join(agent.keywords[:10])}
- Capabilities: {', '.join(agent.capabilities)}
""")
return f"""You are an intelligent task decomposer for a multi-agent system. Your job is to analyze user queries and route them to the most appropriate specialized agents.
AVAILABLE AGENTS:
{chr(10).join(agent_descriptions)}
TASK DECOMPOSITION RULES:
1. **Analyze Intent**: Identify the primary purpose of the user's query
2. **Extract Entities**: Find locations, keywords, parameters, and specific requirements
3. **Route Intelligently**: Choose the most appropriate agent(s) based on intent and entities
4. **Handle Complex Queries**: Break down multi-intent queries into separate tasks
5. **Provide Fallbacks**: Use web_agent for ambiguous or unsupported queries
RESPONSE FORMAT:
Always respond with valid JSON in this exact format:
{{
"analysis": {{
"query_type": "simple|complex|ambiguous",
"primary_intent": "brief description of main intent",
"complexity_score": 0.0-1.0,
"location_extracted": "location if found or null",
"entities": ["entity1", "entity2"],
"reasoning": "brief explanation of your analysis"
}},
"tasks": [
{{
"task_id": "unique_id",
"agent_id": "agent_name",
"description": "clear task description",
"parameters": {{"param1": "value1"}},
"confidence": 0.0-1.0,
"priority": 1-5
}}
]
}}
TOOL PARAMETER SPECIFICATIONS:
- **web_search**: {{"query": "search_terms", "max_results": 5}}
- **sentiment_analysis**: {{"text": "text_to_analyze"}}
- **place_search**: {{"query": "location", "max_distance": 20}}
- **restaurant_search**: {{"query": "location", "cuisine": "cuisine_type_or_null"}}
- **hiking_search**: {{"location": "location", "difficulty": "easy|moderate|hard|null", "max_distance": 30}}
COMPREHENSIVE EXAMPLES:
**Financial/Stock Queries (Enhanced with Ticker Detection):**
Query: "What's NVIDIA's current stock price?"
{{
"analysis": {{"query_type": "simple", "primary_intent": "get financial data", "complexity_score": 0.3, "location_extracted": null, "entities": ["NVIDIA", "stock price"], "reasoning": "Financial query for real-time stock data - ticker detection will enhance this"}},
"tasks": [{{"task_id": "web_001", "agent_id": "web_agent", "description": "Get current NVIDIA stock price with intelligent ticker detection", "parameters": {{"query": "NVIDIA stock price", "max_results": 5}}, "confidence": 0.95, "priority": 1}}]
}}
**General Web Search:**
Query: "Latest news about AI technology"
{{
"analysis": {{"query_type": "simple", "primary_intent": "search for news", "complexity_score": 0.3, "location_extracted": null, "entities": ["news", "AI", "technology"], "reasoning": "General web search for current information"}},
"tasks": [{{"task_id": "web_002", "agent_id": "web_agent", "description": "Search for latest AI technology news", "parameters": {{"query": "latest AI technology news", "max_results": 5}}, "confidence": 0.9, "priority": 1}}]
}}
**Hiking/Outdoor Queries:**
Query: "Find moderate hiking trails near Seattle within 30 miles"
{{
"analysis": {{"query_type": "simple", "primary_intent": "find hiking trails", "complexity_score": 0.4, "location_extracted": "Seattle", "entities": ["hiking", "trails", "moderate", "Seattle", "30 miles"], "reasoning": "Outdoor activity search with specific location and difficulty"}},
"tasks": [{{"task_id": "hiking_001", "agent_id": "hiking_agent", "description": "Find moderate hiking trails near Seattle", "parameters": {{"location": "Seattle", "difficulty": "moderate", "max_distance": 30}}, "confidence": 0.95, "priority": 1}}]
}}
**Hotel/Accommodation Queries:**
Query: "Best luxury hotels in Paris near Eiffel Tower"
{{
"analysis": {{"query_type": "simple", "primary_intent": "find accommodation", "complexity_score": 0.4, "location_extracted": "Paris", "entities": ["hotels", "luxury", "Paris", "Eiffel Tower"], "reasoning": "Accommodation search with location and luxury preference"}},
"tasks": [{{"task_id": "place_001", "agent_id": "location_agent", "description": "Find luxury hotels in Paris near Eiffel Tower", "parameters": {{"query": "luxury hotels Paris near Eiffel Tower", "max_distance": 20}}, "confidence": 0.9, "priority": 1}}]
}}
**Restaurant/Food Queries:**
Query: "Italian restaurants in New York with outdoor seating"
{{
"analysis": {{"query_type": "simple", "primary_intent": "find restaurants", "complexity_score": 0.4, "location_extracted": "New York", "entities": ["Italian", "restaurants", "New York", "outdoor seating"], "reasoning": "Restaurant search with cuisine and location preferences"}},
"tasks": [{{"task_id": "rest_001", "agent_id": "restaurant_agent", "description": "Find Italian restaurants in New York with outdoor seating", "parameters": {{"query": "New York", "cuisine": "Italian"}}, "confidence": 0.9, "priority": 1}}]
}}
**Sentiment Analysis Queries:**
Query: "Analyze sentiment: 'This product is amazing and exceeded my expectations!'"
{{
"analysis": {{"query_type": "simple", "primary_intent": "analyze text sentiment", "complexity_score": 0.2, "location_extracted": null, "entities": ["sentiment", "text analysis"], "reasoning": "Clear sentiment analysis request with provided text"}},
"tasks": [{{"task_id": "sent_001", "agent_id": "sentiment_agent", "description": "Analyze sentiment of product review", "parameters": {{"text": "This product is amazing and exceeded my expectations!"}}, "confidence": 0.95, "priority": 1}}]
}}
**Complex Multi-Intent Queries:**
Query: "I'm planning a trip to Tokyo - need hotels and restaurants"
{{
"analysis": {{"query_type": "complex", "primary_intent": "travel planning with accommodation and dining", "complexity_score": 0.7, "location_extracted": "Tokyo", "entities": ["trip", "Tokyo", "hotels", "restaurants"], "reasoning": "Multi-intent travel query requiring both accommodation and restaurant search"}},
"tasks": [
{{"task_id": "place_001", "agent_id": "location_agent", "description": "Find hotels in Tokyo", "parameters": {{"query": "Tokyo", "max_distance": 20}}, "confidence": 0.9, "priority": 1}},
{{"task_id": "rest_001", "agent_id": "restaurant_agent", "description": "Find restaurants in Tokyo", "parameters": {{"query": "Tokyo", "cuisine": null}}, "confidence": 0.9, "priority": 1}}
]
}}
**Weather/News/General Web Queries:**
Query: "Latest news about artificial intelligence developments"
{{
"analysis": {{"query_type": "simple", "primary_intent": "get current news information", "complexity_score": 0.3, "location_extracted": null, "entities": ["news", "artificial intelligence"], "reasoning": "Information retrieval query requiring web search"}},
"tasks": [{{"task_id": "web_001", "agent_id": "web_agent", "description": "Get latest AI news", "parameters": {{"query": "latest news artificial intelligence developments"}}, "confidence": 0.9, "priority": 1}}]
}}
**Ambiguous Queries:**
Query: "Tell me about Paris"
{{
"analysis": {{"query_type": "ambiguous", "primary_intent": "get general information", "complexity_score": 0.5, "location_extracted": "Paris", "entities": ["Paris"], "reasoning": "Vague query - could be travel, history, or general info - use web search"}},
"tasks": [{{"task_id": "web_001", "agent_id": "web_agent", "description": "Get general information about Paris", "parameters": {{"query": "Paris information travel guide"}}, "confidence": 0.7, "priority": 1}}]
}}
INTELLIGENT ROUTING GUIDELINES:
- **Keywords for hiking_agent**: hiking, trails, trek, outdoor, mountain, nature, walk, climbing, adventure
- **Keywords for location_agent**: hotel, hotels, accommodation, stay, lodging, motel, resort, inn
- **Keywords for restaurant_agent**: restaurant, food, dining, eat, cuisine, meal, cafe, bar
- **Keywords for sentiment_agent**: sentiment, analyze, opinion, feeling, emotion, review, mood
- **Keywords for web_agent**: news, weather, stock, price, latest, current, information, what is
PARAMETER EXTRACTION RULES:
- **Locations**: Look for city names, landmarks, "in", "near", "at", "around"
- **Difficulties**: easy, moderate, hard, difficult, challenging, extreme
- **Distances**: "within X miles", "X km radius", "close to"
- **Cuisines**: Italian, Chinese, Mexican, etc.
- **Accommodations**: luxury, budget, boutique, business, etc.
IMPORTANT:
- Always provide valid JSON
- Use exact agent_id values from the list above
- Extract locations and parameters accurately
- Assign appropriate confidence scores based on query clarity
- For unclear queries, use web_agent as fallback
- Be specific in task descriptions and reasoning"""
def _parse_llm_response(self, response, original_query: str) -> Dict[str, Any]:
"""Parse LLM response and extract structured analysis."""
try:
# Handle different response types
if hasattr(response, 'content'):
# ChatMessage object - extract content
response_text = response.content
elif hasattr(response, 'text'):
# Some other response object with text attribute
response_text = response.text
elif isinstance(response, str):
# Already a string
response_text = response
else:
# Try to convert to string
response_text = str(response)
print(f"🔍 Raw response text: {response_text[:500]}...")
# Clean up markdown code blocks if present
import re
# Remove markdown code block markers
response_text = re.sub(r'```json\s*', '', response_text)
response_text = re.sub(r'```\s*$', '', response_text)
response_text = response_text.strip()
# Try to extract JSON from the response - more robust pattern
json_patterns = [
r'\{[^{}]*(?:\{[^{}]*\}[^{}]*)*\}', # Simple nested braces
r'\{.*\}', # Original fallback pattern
]
analysis = None
# First try: Direct JSON parsing if the response looks like pure JSON
if response_text.strip().startswith('{') and response_text.strip().endswith('}'):
try:
analysis = json.loads(response_text.strip())
print(f"✅ Successfully parsed JSON via direct parsing")
except json.JSONDecodeError:
print(f"⚠️ Direct JSON parsing failed, trying pattern matching")
# Second try: Pattern matching
if not analysis:
for pattern in json_patterns:
json_match = re.search(pattern, response_text, re.DOTALL)
if json_match:
try:
json_text = json_match.group().strip()
print(f"🔍 Extracted JSON: {json_text[:200]}...")
analysis = json.loads(json_text)
print(f"✅ Successfully parsed JSON analysis via pattern matching")
break
except json.JSONDecodeError as json_error:
print(f"❌ JSON decode error with pattern {pattern}: {json_error}")
continue
# Third try: Find balanced braces manually
if not analysis:
try:
analysis = self._extract_json_with_balanced_braces(response_text)
if analysis:
print(f"✅ Successfully parsed JSON via balanced brace extraction")
except Exception as brace_error:
print(f"❌ Balanced brace extraction failed: {brace_error}")
if analysis:
return analysis
else:
raise ValueError("No valid JSON found in response")
except Exception as e:
print(f"❌ Error parsing LLM response: {e}")
print(f"🔄 Falling back to keyword-based analysis")
# Return fallback analysis
return {
"analysis": {
"query_type": "simple",
"primary_intent": "general query",
"complexity_score": 0.5,
"location_extracted": None,
"entities": [],
"reasoning": f"LLM parsing failed: {str(e)}"
},
"tasks": [
{
"task_id": "web_fallback",
"agent_id": "web_agent",
"description": original_query,
"parameters": {"query": original_query, "category": None},
"confidence": 0.5,
"priority": 1
}
]
}
def _extract_json_with_balanced_braces(self, text: str) -> Optional[Dict[str, Any]]:
"""Extract JSON by finding balanced braces manually."""
import json
# Find the first opening brace
start_idx = text.find('{')
if start_idx == -1:
return None
# Count braces to find the matching closing brace
brace_count = 0
end_idx = start_idx
in_string = False
escape_next = False
for i, char in enumerate(text[start_idx:], start_idx):
if escape_next:
escape_next = False
continue
if char == '\\':
escape_next = True
continue
if char == '"' and not escape_next:
in_string = not in_string
continue
if not in_string:
if char == '{':
brace_count += 1
elif char == '}':
brace_count -= 1
if brace_count == 0:
end_idx = i
break
if brace_count == 0:
json_text = text[start_idx:end_idx + 1]
try:
return json.loads(json_text)
except json.JSONDecodeError:
return None
return None
def _create_subtasks(self, analysis: Dict[str, Any], original_query: str) -> List[SubTask]:
"""Convert LLM analysis into SubTask objects."""
subtasks = []
tasks = analysis.get("tasks", [])
if not tasks:
# Fallback if no tasks generated
tasks = [{
"task_id": "fallback_001",
"agent_id": "web_agent",
"description": original_query,
"parameters": {"query": original_query},
"confidence": 0.5,
"priority": 1
}]
for task_data in tasks:
agent_id = task_data.get("agent_id", "web_agent")
# Map agent_id to task_type
task_type_mapping = {
"sentiment_agent": TaskType.SENTIMENT_ANALYSIS,
"location_agent": TaskType.LOCATION_SEARCH,
"restaurant_agent": TaskType.RESTAURANT_SEARCH,
"hiking_agent": TaskType.HIKING_SEARCH,
"web_agent": TaskType.WEB_SEARCH
}
task_type = task_type_mapping.get(agent_id, TaskType.WEB_SEARCH)
subtask = SubTask(
id=task_data.get("task_id", str(uuid.uuid4())),
task_type=task_type,
description=task_data.get("description", original_query),
parameters=task_data.get("parameters", {"query": original_query}),
agent_id=agent_id,
confidence=task_data.get("confidence", 0.5)
)
subtasks.append(subtask)
return subtasks
def _fallback_decomposition(self, user_query: str) -> List[SubTask]:
"""Fallback decomposition using simple keyword matching."""
print(f"🔄 Using fallback decomposition for: '{user_query}'")
query_lower = user_query.lower()
# Simple keyword-based classification
if any(word in query_lower for word in ["sentiment", "feeling", "opinion", "emotion", "analyze"]):
print(f"🎭 Detected sentiment analysis request")
# Extract text to analyze - look for text in quotes or after "analyze sentiment:"
import re
text_to_analyze = user_query
# Try to extract quoted text first
quote_pattern = r"['\"]([^'\"]+)['\"]"
quote_match = re.search(quote_pattern, user_query)
if quote_match:
text_to_analyze = quote_match.group(1)
print(f"📝 Extracted quoted text: '{text_to_analyze}'")
# Try to extract text after "analyze sentiment:" or similar patterns
elif "analyze sentiment:" in query_lower:
parts = user_query.split(":", 1)
if len(parts) > 1:
text_to_analyze = parts[1].strip().strip("'\"")
print(f"📝 Extracted text after colon: '{text_to_analyze}'")
# Try to extract text after "sentiment" keyword
elif "sentiment" in query_lower:
# Look for patterns like "sentiment of X" or "analyze X sentiment"
sentiment_patterns = [
r"sentiment[:\s]+['\"]?([^'\"]+)['\"]?",
r"analyze[:\s]+['\"]?([^'\"]+)['\"]?\s+sentiment",
r"['\"]([^'\"]+)['\"].*sentiment"
]
for pattern in sentiment_patterns:
match = re.search(pattern, user_query, re.IGNORECASE)
if match:
text_to_analyze = match.group(1).strip()
print(f"📝 Extracted text via pattern: '{text_to_analyze}'")
break
print(f"🎯 Final text for sentiment analysis: '{text_to_analyze}'")
return [SubTask(
id=str(uuid.uuid4()),
task_type=TaskType.SENTIMENT_ANALYSIS,
description=f"Analyze sentiment: {text_to_analyze}",
parameters={"text": text_to_analyze},
agent_id="sentiment_agent",
confidence=0.8
)]
elif any(word in query_lower for word in ["hiking", "trail", "trails", "trek", "trekking", "hike", "hikes"]):
# Extract location and difficulty for hiking
import re
# Extract location patterns
location_patterns = [
r"(?:in|at|near|around|close to)\s+([a-zA-Z\s,]+?)(?:\s+within|\s+\d|$|\.|,)",
r"([A-Z][a-zA-Z\s]+?)(?:\s+within|\s+\d|$)"
]
location = None
for pattern in location_patterns:
location_match = re.search(pattern, user_query)
if location_match:
location = location_match.group(1).strip()
break
if not location:
location = user_query # Fallback to full query
# Extract difficulty
difficulty = None
if "easy" in query_lower:
difficulty = "easy"
elif "moderate" in query_lower:
difficulty = "moderate"
elif any(word in query_lower for word in ["hard", "difficult", "challenging"]):
difficulty = "hard"
elif any(word in query_lower for word in ["very hard", "extreme", "strenuous"]):
difficulty = "very_hard"
# Extract distance
max_distance = 30 # Default
distance_match = re.search(r"within\s+(\d+)\s*(?:mile|miles|mi)", query_lower)
if distance_match:
max_distance = int(distance_match.group(1))
return [SubTask(
id=str(uuid.uuid4()),
task_type=TaskType.HIKING_SEARCH,
description=f"Find hiking trails: {user_query}",
parameters={"location": location, "difficulty": difficulty, "max_distance": max_distance},
agent_id="hiking_agent",
confidence=0.8
)]
elif any(word in query_lower for word in ["hotel", "accommodation", "stay", "place"]):
return [SubTask(
id=str(uuid.uuid4()),
task_type=TaskType.LOCATION_SEARCH,
description=f"Find accommodations: {user_query}",
parameters={"query": user_query, "max_distance": 20},
agent_id="location_agent",
confidence=0.7
)]
elif any(word in query_lower for word in ["restaurant", "food", "dining", "eat"]):
return [SubTask(
id=str(uuid.uuid4()),
task_type=TaskType.RESTAURANT_SEARCH,
description=f"Find restaurants: {user_query}",
parameters={"query": user_query, "cuisine": None},
agent_id="restaurant_agent",
confidence=0.7
)]
else:
# Default to web search
return [SubTask(
id=str(uuid.uuid4()),
task_type=TaskType.WEB_SEARCH,
description=f"Web search: {user_query}",
parameters={"query": user_query, "category": None},
agent_id="web_agent",
confidence=0.6
)]
async def test_ticker_detection(self, test_queries: List[str] = None) -> Dict[str, str]:
"""Test ticker detection on various queries to help debug issues."""
if test_queries is None:
test_queries = [
"What's the current stock price of NVDA?",
"NVDA stock price",
"Get NVIDIA stock price",
"What is TSLA trading at?",
"Apple stock price",
"AAPL current price"
]
results = {}
print("🧪 Testing ticker detection...")
for query in test_queries:
detected = await self.detect_ticker_symbol(query)
results[query] = detected
print(f" '{query}' → '{detected}'")
return results
async def detect_ticker_symbol(self, user_query: str) -> str:
"""
Use LLM to detect and extract ticker symbols from financial queries.
"""
try:
# First check for obvious ticker symbols in the query
import re
print(f"🔍 Analyzing query for ticker: '{user_query}'")
# Common ticker patterns - improved to catch more cases
ticker_patterns = [
r'\b([A-Z]{1,5})\b(?:\s+stock|\s+price|\s+quote)', # NVDA stock, AAPL price
r'\bof\s+([A-Z]{2,5})\b', # "price of NVDA"
r'\b([A-Z]{2,5})\s*\??\s*$', # NVDA at end of query
r'\b([A-Z]{2,5})\b(?=\s)', # Standalone uppercase 2-5 letters followed by space
r'\b([A-Z]{2,5})\b', # Any 2-5 letter uppercase sequence
]
# Known ticker mappings for common companies
company_tickers = {
'nvidia': 'NVDA',
'apple': 'AAPL',
'tesla': 'TSLA',
'microsoft': 'MSFT',
'google': 'GOOGL',
'amazon': 'AMZN',
'meta': 'META',
'facebook': 'META',
'spy': 'SPY',
'qqq': 'QQQ'
}
query_lower = user_query.lower()
# Check for direct ticker matches first
for i, pattern in enumerate(ticker_patterns):
matches = re.findall(pattern, user_query, re.IGNORECASE)
print(f" Pattern {i+1} ('{pattern}'): {matches}")
for match in matches:
if len(match) >= 2 and match.upper() not in ['THE', 'AND', 'FOR', 'ARE', 'BUT', 'NOT', 'YOU', 'ALL', 'CAN', 'HER', 'WAS', 'ONE', 'OUR', 'HAD', 'BUT', 'WHAT', 'BEEN', 'THAT', 'WITH', 'THIS']:
print(f"🎯 Direct ticker pattern match found: {match.upper()}")
return match.upper()
# Check for company name matches
for company, ticker in company_tickers.items():
if company in query_lower:
print(f"🏢 Company name match found: {company} → {ticker}")
return ticker
# Use LLM as fallback for complex cases
prompt = f"""
You are a financial assistant. Your task is to identify stock ticker symbols in queries.
Query: "{user_query}"
If this query mentions a company or stock, return ONLY the ticker symbol (e.g., "AAPL", "TSLA", "NVDA").
If no ticker can be identified, return "UNKNOWN".
Examples:
- "Apple stock price" → AAPL
- "Tesla earnings" → TSLA
- "NVIDIA performance" → NVDA
- "Microsoft news" → MSFT
- "What's the current stock price of NVDA?" → NVDA
- "weather forecast" → UNKNOWN
Response (ticker only):"""
if self.model_name == "anthropic" and self.model:
response = self.model([{"role": "user", "content": prompt}])
ticker = str(response).strip().upper()
# Validate ticker format
if ticker and ticker != "UNKNOWN" and len(ticker) <= 5 and ticker.isalpha():
print(f"🤖 LLM ticker detection: {ticker}")
return ticker
print(f"❓ No ticker detected for query: {user_query}")
return "UNKNOWN"
except Exception as e:
print(f"❌ Ticker detection failed: {e}")
return "UNKNOWN"
async def enhance_financial_query(self, user_query: str) -> str:
"""Enhance financial queries with ticker symbol detection."""
ticker_result = await self.detect_ticker_symbol(user_query)
if ticker_result != "UNKNOWN":
# Specific ticker found - create focused financial query
enhanced_query = f"{ticker_result} stock price quote market data"
print(f"🎯 Enhanced financial query: '{user_query}' → '{enhanced_query}' (ticker: {ticker_result})")
return enhanced_query
else:
print(f"🔍 Using original query: '{user_query}' (no ticker detected)")
return user_query
class MCPClientManager:
"""Enhanced MCP Client Manager with LLM-powered task decomposition."""
def __init__(self, server_url: str = "http://localhost:7861/gradio_api/mcp/sse", model_name: str = "anthropic"):
self.server_url = server_url
self.model_name = model_name
self.task_decomposer = LLMTaskDecomposer(model_name)
self.mcp_client = None
self.session_id = str(uuid.uuid4())
self.is_connected = False
self.available_tools = []
async def connect_to_server(self) -> bool:
"""Connect to an already running MCP server using smolagents MCPClient."""
max_retries = 3 if "localhost" in self.server_url else 2
retry_delay = 2
for attempt in range(max_retries):
try:
if attempt > 0:
print(f"🔄 Retry attempt {attempt + 1}/{max_retries}")
await asyncio.sleep(retry_delay)
print(f"🔗 Attempting to connect to MCP server at: {self.server_url}")
# Add timeout for remote connections
timeout = 15 if "localhost" in self.server_url else 45
# Create MCP client using smolagents with explicit transport
self.mcp_client = MCPClient({
"url": self.server_url,
"transport": "sse", # Explicitly specify SSE transport
"timeout": timeout
})
# Get available tools with timeout
try:
print("🔍 Fetching available tools...")
self.available_tools = self.mcp_client.get_tools()
tool_names = [tool.name for tool in self.available_tools]
print(f"✅ Connected to MCP server. Available tools: {tool_names}")
# Debug: Print detailed tool information
if self.available_tools:
print("📋 Tool Details:")
for tool in self.available_tools:
print(f" • {tool.name}")
else:
print("⚠️ Warning: No tools found on the server")
# Check if tools have prefixes and suggest mapping
if tool_names and any("_" in name for name in tool_names):
print("🔧 Detected prefixed tool names - using flexible matching")
except Exception as tools_error:
print(f"⚠️ Warning: Connected but failed to get tools: {tools_error}")
self.available_tools = []
self.is_connected = True
return True
except Exception as e:
error_msg = str(e)
if "timeout" in error_msg.lower() or "connection" in error_msg.lower():
print(f"⏱️ Connection attempt {attempt + 1} failed: {error_msg}")
else:
print(f"❌ Connection attempt {attempt + 1} failed: {error_msg}")
if attempt == max_retries - 1:
print(f"❌ Failed to connect to MCP server after {max_retries} attempts")
print(f"💡 Connection troubleshooting for: {self.server_url}")
if "localhost" in self.server_url:
print("🏠 LOCAL SERVER ISSUES:")
print(" • Make sure the MCP server is running locally")
print(" • Check if port 7861 is available")
print(" • Try running: python server.py in the mcp_server directory")
else:
print("🌐 REMOTE SERVER ISSUES:")
if "hf.space" in self.server_url:
print(" • The Hugging Face Space might be PRIVATE (not publicly accessible)")
print(" • Make the Space PUBLIC in HF settings, or")
print(" • Use a local server instead")
print(" • Check your internet connection")
print(" • Verify the server URL is correct and accessible")
await self._cleanup()
return False
async def execute_subtask(self, subtask: SubTask) -> Dict[str, Any]:
"""Execute a subtask using the MCP tool."""
if not self.is_connected or not self.mcp_client:
return {"error": "Not connected to MCP server"}
try:
agent = self.task_decomposer.agents.get(subtask.agent_id)
if not agent:
return {"error": f"Agent {subtask.agent_id} not found"}
tool_name = agent.tool_name
# Debug: print available tools
available_tool_names = [tool.name for tool in self.available_tools]
print(f"🔍 Looking for tool '{tool_name}' among available tools: {available_tool_names}")
# Find the tool - support both exact matches and suffix matches (for prefixed tools)
tool = None
for available_tool in self.available_tools:
# First try exact match
if available_tool.name == tool_name:
tool = available_tool
print(f"✅ Found exact match: {available_tool.name}")
break
# Then try suffix match (for tools like "test_mcp_server_sentiment_analysis")
elif available_tool.name.endswith(f"_{tool_name}") or available_tool.name.endswith(tool_name):
tool = available_tool
print(f"✅ Found suffix match: {available_tool.name} matches {tool_name}")
break
if not tool:
return {
"error": f"Tool {tool_name} not available on server",
"available_tools": available_tool_names,
"requested_tool": tool_name,
"agent_id": subtask.agent_id
}
# Special handling for web search with ticker detection
if tool_name == "web_search" and subtask.agent_id == "web_agent":
# Enhance query with ticker detection for financial queries
original_query = subtask.parameters.get("query", "")
enhanced_query = await self.task_decomposer.enhance_financial_query(original_query)
subtask.parameters["query"] = enhanced_query
print(f"💡 Web search query enhanced: '{original_query}' → '{enhanced_query}'")
# Map and filter parameters based on tool type
filtered_params = self._filter_tool_parameters(tool_name, subtask.parameters)
print(f"🔧 Filtered parameters for {tool_name}: {filtered_params}")
# Execute the tool
try:
result = tool(**filtered_params)
# Handle the result - parse JSON string if needed
if isinstance(result, str):
try:
parsed_result = json.loads(result)
except json.JSONDecodeError:
parsed_result = {"result": result}
elif isinstance(result, dict):
parsed_result = result
elif hasattr(result, 'content'):
# If it's a tool result object
parsed_result = {"result": str(result.content)}
else:
parsed_result = {"result": str(result)}
subtask.status = "completed"
subtask.result = parsed_result
return parsed_result
except Exception as tool_error:
return {"error": f"Tool execution error: {str(tool_error)}", "tool_name": tool_name}
except Exception as e:
subtask.status = "failed"
return {"error": f"Subtask execution failed: {str(e)}", "subtask_id": subtask.id}
def _filter_tool_parameters(self, tool_name: str, parameters: Dict[str, Any]) -> Dict[str, Any]:
"""Filter and map parameters based on tool requirements."""
# Parameter mappings for each tool
tool_param_mappings = {
"web_search": { # Updated to handle actual web_search tool
"allowed_params": ["query", "max_results"],
"param_mapping": {
"search_query": "query",
"search_term": "query",
"q": "query",
"data_type": None, # Remove this parameter
"category": None # Remove this parameter
}
},
"sentiment_analysis": {
"allowed_params": ["text"],
"param_mapping": {
"input_text": "text",
"content": "text"
}
},
"place_search": {
"allowed_params": ["query", "max_distance"],
"param_mapping": {
"location": "query",
"search_query": "query",
"distance": "max_distance"
}
},
"restaurant_search": {
"allowed_params": ["query", "cuisine"],
"param_mapping": {
"location": "query",
"search_query": "query",
"cuisine_type": "cuisine"
}
},
"hiking_search": {
"allowed_params": ["location", "difficulty", "max_distance"],
"param_mapping": {
"query": "location",
"search_query": "location",
"skill_level": "difficulty"
}
}
}
mapping_config = tool_param_mappings.get(tool_name, {
"allowed_params": ["query"],
"param_mapping": {}
})
filtered_params = {}
for param_key, param_value in parameters.items():
# Check if parameter should be mapped to a different name
mapped_key = mapping_config["param_mapping"].get(param_key, param_key)
# Skip parameters that are mapped to None (should be removed)
if mapped_key is None:
continue
# Skip None or empty values
if param_value is None or param_value == "":
continue
# Only include allowed parameters
if mapped_key in mapping_config["allowed_params"]:
filtered_params[mapped_key] = param_value
# Ensure required parameters exist with defaults
if tool_name in ["web_search"] and "query" not in filtered_params:
# If no query parameter, use the first available parameter value
if parameters:
filtered_params["query"] = str(list(parameters.values())[0])
elif tool_name == "hiking_search" and "location" not in filtered_params:
# For hiking, ensure location is provided
if "query" in parameters and parameters["query"]:
filtered_params["location"] = str(parameters["query"])
elif tool_name == "restaurant_search" and "query" not in filtered_params:
# For restaurants, ensure query is provided
if "location" in parameters and parameters["location"]:
filtered_params["query"] = str(parameters["location"])
elif tool_name == "place_search" and "query" not in filtered_params:
# For places, ensure query is provided
if "location" in parameters and parameters["location"]:
filtered_params["query"] = str(parameters["location"])
return filtered_params
def test_tool_matching(self) -> str:
"""Test tool matching logic for debugging purposes."""
if not self.available_tools:
return "❌ No tools available to test"
results = []
results.append("🧪 Tool Matching Test Results:")
results.append("")
# Test each agent's tool against available tools
for agent_id, agent in self.task_decomposer.agents.items():
tool_name = agent.tool_name
results.append(f"🔍 Testing agent '{agent_id}' looking for tool '{tool_name}':")
# Test exact match
exact_match = None
suffix_match = None
for available_tool in self.available_tools:
if available_tool.name == tool_name:
exact_match = available_tool.name
break
elif available_tool.name.endswith(f"_{tool_name}") or available_tool.name.endswith(tool_name):
suffix_match = available_tool.name
if exact_match:
results.append(f" ✅ Exact match found: {exact_match}")
elif suffix_match:
results.append(f" ✅ Suffix match found: {suffix_match}")
else:
results.append(f" ❌ No match found")
results.append("")
return "\n".join(results)
async def process_query(self, user_query: str) -> Dict[str, Any]:
"""Process user query through LLM-powered task decomposition and agent routing."""
try:
# Ensure connection
if not self.is_connected:
success = await self.connect_to_server()
if not success:
return {
"query": user_query,
"error": "Failed to connect to MCP server. Please ensure the server is running separately.",
"status": "failed"
}
# Use LLM to decompose query into subtasks
subtasks = await self.task_decomposer.decompose_query(user_query)
# Execute subtasks
results = []
for subtask in subtasks:
result = await self.execute_subtask(subtask)
results.append({
"subtask_id": subtask.id,
"task_type": subtask.task_type.value,
"agent": subtask.agent_id,
"description": subtask.description,
"confidence": subtask.confidence,
"result": result
})
# Aggregate results
response = {
"query": user_query,
"subtasks_count": len(subtasks),
"subtasks": results,
"status": "completed",
"summary": self._generate_summary(user_query, results)
}
return response
except Exception as e:
print(f"❌ Failed to process query: {e}")
return {
"query": user_query,
"error": f"Query processing failed: {str(e)}",
"status": "failed"
}
def _generate_summary(self, query: str, results: List[Dict[str, Any]]) -> str:
"""Generate summary of all subtask results - now simplified since server handles formatting."""
try:
if not results:
return f"# 🤔 No Results\n\nNo results available for your query. Please try a different search term.\n\n---\n*🧠 Powered by AI Task Decomposition*"
# Check if we have a pre-formatted summary from server
for result in results:
try:
if isinstance(result, dict) and isinstance(result.get("result"), dict):
# Check for direct summary first
if result["result"].get("summary"):
return result["result"]["summary"]
# Check for formatted result content
elif result["result"].get("result"):
formatted_content = result["result"]["result"]
if isinstance(formatted_content, str) and "🎭" in formatted_content:
# This is pre-formatted content from server - return it directly
return formatted_content
except (KeyError, TypeError, AttributeError) as e:
print(f"⚠️ Warning: Error accessing result summary: {e}")
continue
# Generate custom formatted summary
summary_parts = [f"# 🎯 Results for: *{query}*", ""]
for i, result in enumerate(results, 1):
try:
task_type = result.get("task_type", "unknown")
description = result.get("description", "No description")
confidence = result.get("confidence", 0.5)
result_data = result.get("result", {})
if isinstance(result_data, dict) and "error" not in result_data:
summary_parts.append(f"## 🌐 Task {i}: {task_type.replace('_', ' ').title()}")
summary_parts.append(f"**Confidence:** {confidence:.1%}")
summary_parts.append("")
# Extract and format the actual content
if result_data.get("summary"):
# Direct summary
summary_parts.append(result_data["summary"])
elif result_data.get("result"):
# Extract formatted content from nested result
content = result_data["result"]
if isinstance(content, str):
# Clean up any escaped newlines and display formatted content
formatted_content = content.replace('\\n', '\n').replace('\\t', '\t')
summary_parts.append(formatted_content)
else:
# Handle other data types
summary_parts.append(self._format_result_content(content))
else:
# Fallback - format the entire result_data
summary_parts.append(self._format_result_content(result_data))
summary_parts.append("")
else:
# Handle errors
summary_parts.append(f"## ❌ {task_type.replace('_', ' ').title()} - Error")
if isinstance(result_data, dict):
error_msg = result_data.get('error', 'Unknown error occurred')
else:
error_msg = str(result_data)
summary_parts.append(f"**Issue:** {error_msg}")
summary_parts.append("")
except Exception as e:
print(f"⚠️ Warning: Error processing result {i}: {e}")
summary_parts.append(f"## ❌ Task {i} - Processing Error")
summary_parts.append(f"**Issue:** {str(e)}")
summary_parts.append("")
summary_parts.append("---")
return "\n".join(summary_parts)
except Exception as e:
print(f"❌ Error in _generate_summary: {e}")
return f"# ❌ Summary Generation Error\n\nFailed to generate summary: {str(e)}\n\n---\n*🧠 Powered by AI Task Decomposition*"
async def _cleanup(self):
"""Clean up resources."""
if self.mcp_client:
try:
self.mcp_client.disconnect()
except:
pass
self.mcp_client = None
self.is_connected = False
async def disconnect(self):
"""Disconnect from MCP server."""
await self._cleanup()
print("🔌 Disconnected from MCP server")
def create_mcp_client_interface(server_url: str = "http://localhost:7861/gradio_api/mcp/sse", model_name: str = "anthropic"):
"""Create the Gradio interface for MCP Client with LLM-powered task decomposition."""
# Create the client manager
async def process_query(query: str):
"""Process user query through MCP with LLM decomposition."""
if not query.strip():
return "Please enter a query to process."
try:
# Create a new client manager for each query to ensure fresh connection
client_manager = MCPClientManager(server_url, model_name)
# Process the query
result = await client_manager.process_query(query)
# Clean up
await client_manager.disconnect()
if result and 'summary' in result:
return result['summary']
else:
return "❌ No results found or error occurred during processing."
except Exception as e:
return f"❌ Error processing query: {str(e)}"
# Check available models and API keys
available_models = []
if ANTHROPIC_API_KEY:
available_models.append(("🤖 Claude Sonnet 4 via Anthropic", "anthropic"))
if not available_models:
available_models.append(("❌ No API Keys Configured", "none"))
# Custom CSS for better UI
css = """
.gradio-container {
max-width: 1200px !important;
margin: auto !important;
}
.header-container {
background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
padding: 2rem;
border-radius: 15px;
margin-bottom: 2rem;
color: white;
text-align: center;
}
.model-info {
background: #f8fafc;
border: 1px solid #e2e8f0;
border-radius: 10px;
padding: 1rem;
margin: 1rem 0;
}
.example-btn {
margin: 0.25rem !important;
background: linear-gradient(45deg, #4f46e5, #7c3aed) !important;
border: none !important;
color: white !important;
}
.example-btn:hover {
transform: translateY(-2px);
box-shadow: 0 4px 12px rgba(79, 70, 229, 0.4) !important;
}
.input-section {
background: #ffffff;
border: 1px solid #e5e7eb;
border-radius: 12px;
padding: 1.5rem;
margin: 1rem 0;
box-shadow: 0 2px 8px rgba(0,0,0,0.05);
}
.results-container {
background: #ffffff;
border: 1px solid #e5e7eb;
border-radius: 12px;
padding: 1.5rem;
margin: 1rem 0;
box-shadow: 0 2px 8px rgba(0,0,0,0.05);
}
.control-buttons {
margin-top: 1rem;
gap: 1rem;
}
.markdown-content {
line-height: 1.6;
}
"""
with gr.Blocks(css=css, title="🚀 VOYAGER AI") as demo:
# Header Section
with gr.Column(elem_classes="header-container"):
gr.HTML("""
<h1 style="margin: 0; font-size: 2.5rem; font-weight: bold;">
🚀 VOYAGER AI
</h1>
<p style="margin: 0.5rem 0 0 0; font-size: 1.2rem; opacity: 0.9;">
Intelligent AI Assistant with Multi-Agent Coordination
</p>
""")
# Main Interface
with gr.Column():
# Examples Section
with gr.Column():
gr.HTML("""
<h3 style="color: #1f2937; margin: 20px 0 15px 0; font-size: 18px; font-weight: 600;">
💡 Quick Start - Try These Examples:
</h3>
""")
with gr.Row():
with gr.Column(scale=1):
sentiment_btn = gr.Button("😊 Analyze Sentiment", elem_classes=["example-btn"])
hiking_btn = gr.Button("🏔️ Hiking trails near Denver", elem_classes=["example-btn"])
with gr.Column(scale=1):
stock_btn = gr.Button("📈 Stock Prices", elem_classes=["example-btn"])
news_btn = gr.Button("📰 Latest News", elem_classes=["example-btn"])
with gr.Column(scale=1):
hotel_btn = gr.Button("🏨 Find Hotels", elem_classes=["example-btn"])
restaurant_btn = gr.Button("🍽️ Find Restaurants", elem_classes=["example-btn"])
# Input Section
with gr.Column(elem_classes="input-section"):
# Query Input
query_input = gr.Textbox(
placeholder="💬 Ask me anything... (e.g., 'What's NVIDIA's stock price?' or 'Find hotels and restaurants in New york')",
lines=3,
label="Your Query",
show_label=False,
container=False
)
# Control Buttons
with gr.Row(elem_classes="control-buttons"):
submit_btn = gr.Button(
"🧠 Analyze & Execute Query",
variant="primary",
size="lg",
scale=2
)
clear_btn = gr.Button(
"🔄 Clear All",
variant="secondary",
size="lg",
scale=1
)
# Results Display
with gr.Column(elem_classes="results-container"):
gr.HTML("""
<h3 style="color: #1f2937; margin: 0 0 20px 0; font-size: 18px; font-weight: 600;">
🎯 AI Results & Analysis
</h3>
""")
results_output = gr.Markdown(
value=f"""**Welcome to VOYAGER AI!** 🧠
**🎯 How it works:**
1. **Try one of the examples above** or type your question naturally
2. **Click "Analyze & Execute"** to get intelligent results
**✨ Features:**
• 🔍 **Smart Query Analysis** - AI understands your intent
• 📋 **Task Decomposition** - Complex queries broken down into subtasks
• 🤖 **Agent Routing** - Specialized agents for different tasks
• ⚡ **Real-time Data** - Live web search and current information
• 🎨 **Professional Results** - Clean, formatted responses
**🚀 Ready to start?** Try one of the example buttons above or type your own query!
---
🤖 **Current Model:** {available_models[0][0] if available_models[0][1] != "none" else "No API Keys Configured"}""",
show_label=False,
container=False,
elem_classes=["markdown-content"]
)
# Event handlers
submit_btn.click(
process_query,
inputs=[query_input],
outputs=[results_output]
)
clear_btn.click(
fn=lambda: ("", """**Interface Cleared!** 🧹
Ready for your next query. Try the example buttons above or ask me anything!
💡 **Quick Tips:**
- Try asking about stock prices, weather, news, or travel
- Use natural language - no need for specific commands
- Complex queries are automatically broken down into tasks"""),
outputs=[query_input, results_output]
)
query_input.submit(
process_query,
inputs=[query_input],
outputs=[results_output]
)
# Example button handlers with better queries
sentiment_btn.click(
fn=lambda: "Analyze sentiment: 'I absolutely love this new AI technology - it's revolutionary and amazing!'",
outputs=query_input
)
hiking_btn.click(
fn=lambda: "Find moderate hiking trails near Denver",
outputs=query_input
)
stock_btn.click(
fn=lambda: "What's the current stock price of SPY?",
outputs=query_input
)
news_btn.click(
fn=lambda: "Latest news about artificial intelligence and technology",
outputs=query_input
)
hotel_btn.click(
fn=lambda: "Find luxury hotels in New york",
outputs=query_input
)
restaurant_btn.click(
fn=lambda: "Best Italian restaurants in New York",
outputs=query_input
)
return demo
async def main():
"""Main entry point for the LLM-Powered MCP Client."""
# Set up environment first
setup_environment()
parser = argparse.ArgumentParser(description="LLM-Powered MCP Client with Intelligent Task Decomposition")
parser.add_argument(
"--server-url",
default="https://agents-mcp-hackathon-test-mcp-server.hf.space/gradio_api/mcp/sse",
help="MCP server URL (default: https://agents-mcp-hackathon-test-mcp-server.hf.space/gradio_api/mcp/sse)"
)
parser.add_argument(
"--local",
action="store_true",
help="Use local MCP server (http://localhost:7860/gradio_api/mcp/sse) instead of remote server"
)
parser.add_argument(
"--port",
type=int,
default=7862,
help="Port to run the client interface (default: 7862)"
)
parser.add_argument(
"--model",
default="anthropic",
choices=["anthropic"],
help="LLM model for task decomposition (default: anthropic)"
)
args = parser.parse_args()
# Override server URL if --local flag is used
if args.local:
server_url = "http://localhost:7860/gradio_api/mcp/sse"
print("🏠 Local development mode enabled")
else:
server_url = args.server_url
print("☁️ Using remote MCP server")
print("🚀 VOYAGER AI - LLM-Powered Task Decomposition")
print("🧠 Intelligent query analysis and agent coordination")
print("🤖 Available Models:")
if ANTHROPIC_API_KEY:
print(" • ✅ Claude Sonnet 4 via Anthropic")
else:
print(" • ❌ Claude Sonnet 4 (API key missing)")
print("")
print(f"📡 MCP Server: {server_url}")
print(f"🧠 Default Model: {args.model}")
print("═" * 50)
# Validate selected model
if args.model == "anthropic" and not ANTHROPIC_API_KEY:
print("⚠️ Warning: Anthropic model selected but API key not configured")
# Create and launch LLM-powered interface
demo = create_mcp_client_interface(server_url, args.model)
print("🌟 Interface ready! Select your model and ask anything naturally!")
# Launch the interface - different configs for local vs deployment
if args.local:
# Local development - with share link
demo.launch(
server_name="0.0.0.0",
server_port=args.port,
share=True,
show_error=True
)
else:
# Production deployment (e.g., Hugging Face Spaces)
demo.launch(
server_name="0.0.0.0",
server_port=7860,
show_error=True
)
if __name__ == "__main__":
try:
asyncio.run(main())
except KeyboardInterrupt:
print("🛑 Client shutdown requested")
except Exception as e:
print(f"❌ Client error: {e}")
sys.exit(1) |