File size: 17,966 Bytes
92c4ae6 | 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 | # -*- coding: utf-8 -*-
"""
Agent Execution Service
Provides centralized agent chat execution with:
- Full governance integration
- WebSocket streaming support
- AgentExecution audit trail
- Episode creation for memory
"""
import logging
import os
import uuid
from datetime import datetime
from typing import Any, Dict, List, Optional
from sqlalchemy.orm import Session
from core.agent_context_resolver import AgentContextResolver
from core.agent_governance_service import AgentGovernanceService
from core.chat_context_manager import get_chat_context_manager
from core.chat_session_manager import get_chat_session_manager
from core.database import get_db_session, SessionLocal
from core.episode_integration import trigger_episode_creation
from core.lancedb_handler import get_chat_history_manager
from core.llm_service import LLMService
from core.models import AgentExecution, AgentInstallation
from core.marketplace_usage_tracker import MarketplaceUsageTracker
from core.personal_budget_service import personal_budget_service
from core.websockets import manager as ws_manager
logger = logging.getLogger(__name__)
class ChatMessage:
"""Simple chat message model"""
def __init__(self, role: str, content: str):
self.role = role
self.content = content
async def execute_agent_chat(
agent_id: str,
message: str,
user_id: str,
session_id: Optional[str] = None,
workspace_id: str = "default",
conversation_history: List[Dict[str, str]] = None,
stream: bool = False
) -> Dict[str, Any]:
"""
Execute agent chat with full governance and streaming support.
This is the centralized service for executing agent chat requests,
used by menubar, mobile, and web platforms.
Args:
agent_id: The ID of the agent to execute
message: User's message to the agent
user_id: User ID making the request
session_id: Optional session ID for conversation continuity
workspace_id: Workspace ID (default for single-tenant)
conversation_history: Optional conversation history for context
stream: Whether to stream response via WebSocket
Returns:
Dictionary containing:
- success: bool
- execution_id: str
- response: str (full response if not streaming)
- agent_id: str
- agent_name: str
- message_id: str (for WebSocket tracking)
- error: str (if failed)
Example:
result = await execute_agent_chat(
agent_id="agent_123",
message="Hello, how can you help me?",
user_id="user_456"
)
print(result["response"])
"""
# Feature flags
governance_enabled = os.getenv("STREAMING_GOVERNANCE_ENABLED", "true").lower() == "true"
emergency_bypass = os.getenv("EMERGENCY_GOVERNANCE_BYPASS", "false").lower() == "true"
agent = None
agent_execution = None
resolution_context = None
governance_check = None
db_session = None
try:
# ============================================
# GOVERNANCE: Agent Resolution & Validation
# ============================================
if governance_enabled and not emergency_bypass:
db_session = SessionLocal()
resolver = AgentContextResolver(db_session)
governance = AgentGovernanceService(db_session)
# Resolve agent for this request
agent, resolution_context = await resolver.resolve_agent_for_request(
user_id=user_id,
session_id=session_id,
requested_agent_id=agent_id,
action_type="chat"
)
if not agent:
logger.warning(f"Agent resolution failed for agent_id={agent_id}, using system default")
# Fall through to system default behavior
# Perform governance check
if agent:
governance_check = governance.can_perform_action(
agent_id=agent.id,
action_type="chat",
require_approval=False
)
if not governance_check.get("allowed", False):
reason = governance_check.get("reason", "Governance policy denied this action")
logger.warning(f"Governance blocked agent chat: {reason}")
return {
"success": False,
"error": f"Action blocked by governance: {reason}",
"agent_id": agent_id,
"execution_id": None
}
# ============================================
# BUDGET: Check Budget (Warning Only, No Blocking)
# ============================================
# Check budget before execution (warning only, does NOT block)
# Personal use = user's responsibility, so we only log warnings
try:
if personal_budget_service.is_budget_exceeded():
logger.warning(
f"Budget exceeded for agent execution (agent_id={agent_id}). "
f"Continuing anyway (personal use = user responsibility)."
)
# Send alert at 100% threshold
personal_budget_service.send_budget_alert(100.0)
else:
# Send alerts at 80% and 90% thresholds
personal_budget_service.send_budget_alert(80.0)
personal_budget_service.send_budget_alert(90.0)
except Exception as budget_error:
logger.error(f"Budget check failed (continuing anyway): {budget_error}")
# Don't block execution on budget check failures
# ============================================
# EXECUTION: Create AgentExecution Record
# ============================================
execution_id = str(uuid.uuid4())
if agent and governance_enabled:
try:
agent_execution = AgentExecution(
id=execution_id,
agent_id=agent.id,
agent_name=agent.name,
agent_category=agent.category,
user_id=user_id,
workspace_id=workspace_id,
session_id=session_id,
action_type="chat",
action_complexity=1,
status="running",
input_data={"message": message},
metadata={
"source": "menubar",
"governance_check": governance_check,
"resolution_context": resolution_context
}
)
if db_session:
db_session.add(agent_execution)
db_session.commit()
db_session.refresh(agent_execution)
except Exception as exec_error:
logger.error(f"Failed to create AgentExecution record: {exec_error}")
# Continue anyway - don't block execution on audit failure
# ============================================
# LLM: Initialize LLM Service
# ============================================
llm_service = LLMService(tenant_id=workspace_id, db=db_session)
# Prepare messages for LLM
messages = []
# Add system message
agent_name = agent.name if agent else "ATOM"
agent_desc = agent.description if agent else "AI Assistant"
messages.append({
"role": "system",
"content": f"""You are {agent_name}, an intelligent AI assistant.
{agent_desc}
Provide helpful, concise responses. Be direct and practical."""
})
# Add conversation history
if conversation_history:
for hist_msg in conversation_history:
messages.append({
"role": hist_msg.get("role", "user"),
"content": hist_msg.get("content", "")
})
# Add current message
messages.append({
"role": "user",
"content": message
})
# Get optimal provider for this request
complexity = llm_service.analyze_query_complexity(message, task_type="chat")
provider_id, model = llm_service.get_optimal_provider(
complexity,
task_type="chat",
prefer_cost=True,
tenant_plan="free",
is_managed_service=False,
requires_tools=False
)
logger.info(f"Executing agent chat with {provider_id}/{model}" +
(f" (agent: {agent.name})" if agent else ""))
# Create unique message ID for WebSocket tracking
message_id = str(uuid.uuid4())
# If streaming is requested, send initial WebSocket message
if stream:
user_channel = f"user:{user_id}"
await ws_manager.broadcast(user_channel, {
"type": "streaming:start",
"id": message_id,
"model": "auto",
"agent_id": agent.id if agent else None,
"agent_name": agent.name if agent else None,
"execution_id": execution_id
})
# Execute chat (streaming or non-streaming)
accumulated_content = ""
tokens_count = 0
start_time = datetime.now()
stream_kwargs = {
"messages": messages,
"model": "auto",
"temperature": 0.7,
"max_tokens": 2000,
"agent_id": agent.id if agent else None
}
# Stream response
# Stream response via LLMService
async for token in llm_service.stream_completion(**stream_kwargs):
accumulated_content += token
tokens_count += 1
# Broadcast token via WebSocket if streaming enabled
if stream:
user_channel = f"user:{user_id}"
await ws_manager.broadcast(user_channel, {
"type": ws_manager.STREAMING_UPDATE,
"id": message_id,
"delta": token,
"complete": False,
"metadata": {
"tokens_so_far": len(accumulated_content),
"execution_id": execution_id
}
})
# Send completion message if streaming
if stream:
user_channel = f"user:{user_id}"
await ws_manager.broadcast(user_channel, {
"type": ws_manager.STREAMING_COMPLETE,
"id": message_id,
"content": accumulated_content,
"complete": True,
"metadata": {
"execution_id": execution_id,
"tokens_total": tokens_count
}
})
# ============================================
# PERSISTENCE: Save to Chat History
# ============================================
try:
chat_history = get_chat_history_manager(workspace_id)
session_manager = get_chat_session_manager(workspace_id)
# Create or use session
if not session_id:
session_id = session_manager.create_session(user_id)
# Save messages
chat_history.add_message(session_id, "user", message)
chat_history.add_message(session_id, "assistant", accumulated_content)
except Exception as persist_error:
logger.error(f"Failed to save chat history: {persist_error}")
# Don't fail the request on persistence errors
# ============================================
# GOVERNANCE: Update Execution Record
# ============================================
if agent_execution and governance_enabled:
try:
end_time = datetime.now()
duration_ms = (end_time - start_time).total_seconds() * 1000
agent_execution.status = "completed"
agent_execution.output_data = {
"response": accumulated_content,
"tokens": tokens_count,
"model": "auto"
}
agent_execution.duration_ms = duration_ms
agent_execution.end_time = end_time
if db_session:
db_session.commit()
# Marketplace Tracking
if agent and agent.type == "marketplace":
try:
installation = db_session.query(AgentInstallation).filter(
AgentInstallation.instantiated_agent_id == agent.id
).first()
if installation:
MarketplaceUsageTracker.track_usage(
item_type="agent",
item_id=installation.template_id,
success=True,
duration_ms=duration_ms
)
except Exception as mt_error:
logger.error(f"Marketplace tracking failed: {mt_error}")
except Exception as update_error:
logger.error(f"Failed to update AgentExecution record: {update_error}")
# Trigger episode creation for memory
try:
await trigger_episode_creation(
user_id=user_id,
agent_id=agent.id if agent else None,
session_id=session_id,
workspace_id=workspace_id
)
except Exception as episode_error:
logger.warning(f"Failed to trigger episode creation: {episode_error}")
# ============================================
# BUDGET: Track Spend After Execution
# ============================================
# Record spend for budget forecasting and tracking
try:
# Estimate cost based on tokens (rough estimation)
# ACU cost: ~$0.0001 per token, API cost varies by provider
estimated_cost = (tokens_count * 0.0001) + 0.001 # Base API call cost
personal_budget_service.record_spend(estimated_cost, execution_id)
except Exception as budget_error:
logger.error(f"Failed to record spend (non-critical): {budget_error}")
# Don't fail execution on budget tracking errors
# Return success
return {
"success": True,
"execution_id": execution_id,
"response": accumulated_content,
"agent_id": agent.id if agent else agent_id,
"agent_name": agent.name if agent else "System",
"message_id": message_id,
"session_id": session_id,
"tokens": tokens_count,
"model": "auto"
}
except Exception as e:
logger.error(f"Agent chat execution failed: {e}", exc_info=True)
# Update execution record as failed
if agent_execution and governance_enabled and db_session:
try:
agent_execution.status = "failed"
agent_execution.error_message = str(e)
agent_execution.end_time = datetime.now()
db_session.commit()
# Marketplace Tracking (Failure)
if agent and agent.type == "marketplace":
try:
installation = db_session.query(AgentInstallation).filter(
AgentInstallation.instantiated_agent_id == agent.id
).first()
if installation:
duration_ms = (datetime.now() - start_time).total_seconds() * 1000
MarketplaceUsageTracker.track_usage(
item_type="agent",
item_id=installation.template_id,
success=False,
duration_ms=duration_ms
)
except Exception as mt_error:
logger.error(f"Marketplace failure tracking failed: {mt_error}")
except Exception as update_error:
logger.error(f"Failed to update failed execution record: {update_error}")
return {
"success": False,
"error": str(e),
"agent_id": agent_id,
"execution_id": execution_id if agent_execution else None
}
finally:
# Clean up database session
if db_session:
try:
db_session.close()
except Exception:
pass
def execute_agent_chat_sync(
agent_id: str,
message: str,
user_id: str,
session_id: Optional[str] = None,
workspace_id: str = "default",
conversation_history: List[Dict[str, str]] = None
) -> Dict[str, Any]:
"""
Synchronous wrapper for execute_agent_chat.
Use this in non-async contexts. This runs the async function in an event loop.
Note: WebSocket streaming is disabled in sync mode.
Args:
Same as execute_agent_chat
Returns:
Same as execute_agent_chat (but without streaming support)
"""
import asyncio
try:
loop = asyncio.get_event_loop()
except RuntimeError:
loop = asyncio.new_event_loop()
asyncio.set_event_loop(loop)
return loop.run_until_complete(
execute_agent_chat(
agent_id=agent_id,
message=message,
user_id=user_id,
session_id=session_id,
workspace_id=workspace_id,
conversation_history=conversation_history,
stream=False # Disable streaming in sync mode
)
)
|