Spaces:
Sleeping
Sleeping
File size: 9,987 Bytes
1bb4678 b86397a 1bb4678 b86397a 1bb4678 |
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 |
"""
Agents Module: Base agent classes and specialized agents.
Supports both Blackboard (free) and Guided (strict) architectures.
Model-agnostic: Works with Gemini, OpenAI, Anthropic, Groq, Ollama, etc.
"""
from abc import ABC, abstractmethod
from dataclasses import dataclass, field
from typing import Any, Dict, List, Optional, Callable
from enum import Enum
from datetime import datetime
import json
import logging
logger = logging.getLogger(__name__)
class AgentRole(Enum):
"""Roles agents can take in the system."""
ARCHITECT = "architect"
BUILDER = "builder"
VALIDATOR = "validator"
OPTIMIZER = "optimizer"
ANALYZER = "analyzer"
SCORER = "scorer"
COORDINATOR = "coordinator"
class AgentState(Enum):
"""Agent execution states."""
IDLE = "idle"
THINKING = "thinking"
EXECUTING = "executing"
WAITING = "waiting"
COMPLETED = "completed"
ERROR = "error"
@dataclass
class AgentContext:
"""Context passed to agents for decision making."""
goal: str
current_circuit: Optional[str] = None
history: List[Dict] = field(default_factory=list)
constraints: Dict = field(default_factory=dict)
shared_data: Dict = field(default_factory=dict)
def add_to_history(self, action: str, result: Any):
self.history.append({
"action": action,
"result": result,
"timestamp": datetime.now().isoformat()
})
@dataclass
class AgentAction:
"""An action an agent wants to take."""
tool_name: str
arguments: Dict
reasoning: str
priority: float = 1.0
@dataclass
class AgentResult:
"""Result of an agent's execution."""
success: bool
data: Any
message: str
actions_taken: List[str] = field(default_factory=list)
execution_time_ms: float = 0.0
class BaseAgent(ABC):
"""
Abstract base class for all agents.
Provides common interface for both Blackboard and Guided architectures.
"""
def __init__(self,
agent_id: str,
role: AgentRole,
tools: List[str] = None,
llm_config: Dict = None):
self.agent_id = agent_id
self.role = role
self.tools = tools or []
self.llm_config = llm_config or {}
self.state = AgentState.IDLE
self.memory: Dict = {}
self._callbacks: List[Callable] = []
@abstractmethod
def decide(self, context: AgentContext) -> Optional[AgentAction]:
"""Decide what action to take given the context."""
pass
@abstractmethod
def execute(self, action: AgentAction, context: AgentContext) -> AgentResult:
"""Execute the decided action."""
pass
def can_handle(self, context: AgentContext) -> bool:
"""Check if this agent can handle the current context."""
return True
def on_state_change(self, callback: Callable):
"""Register callback for state changes."""
self._callbacks.append(callback)
def _set_state(self, new_state: AgentState):
"""Update state and notify callbacks."""
old_state = self.state
self.state = new_state
for cb in self._callbacks:
cb(self.agent_id, old_state, new_state)
def reset(self):
"""Reset agent to initial state."""
self.state = AgentState.IDLE
self.memory.clear()
class LLMAgent(BaseAgent):
"""
Agent that uses an LLM for decision making.
Model-agnostic: Supports Gemini, OpenAI, Anthropic, Groq, Ollama, etc.
Can be used in both Blackboard and Guided modes.
"""
def __init__(self,
agent_id: str,
role: AgentRole,
system_prompt: str,
tools: List[str] = None,
llm_config: Dict = None):
super().__init__(agent_id, role, tools, llm_config)
self.system_prompt = system_prompt
self._adapter = None
def _get_adapter(self):
"""Get the LLM adapter (lazy init)."""
if self._adapter is None:
from config import config
from agents.llm_adapter import get_llm_adapter
# Get API key dynamically (supports HF Spaces Secrets)
api_key = config.llm.get_api_key()
if not api_key:
raise ValueError(
"Missing API key! To use the Google AI API, provide api_key via:\n"
" 1. GOOGLE_API_KEY environment variable (HF Spaces Secrets)\n"
" 2. GENAI_API_KEY environment variable (fallback)\n"
" 3. Set in .env file (local development)"
)
self._adapter = get_llm_adapter(
provider=config.llm.provider,
model=config.llm.model,
api_key=api_key
)
return self._adapter
def _build_messages(self, context: AgentContext) -> List[Dict]:
"""Build message list for LLM."""
messages = [{"role": "system", "content": self.system_prompt}]
context_msg = f"""
Goal: {context.goal}
Current Circuit:
{context.current_circuit or 'None yet'}
Constraints:
{json.dumps(context.constraints, indent=2)}
History (last 5 actions):
{json.dumps(context.history[-5:], indent=2)}
"""
messages.append({"role": "user", "content": context_msg})
return messages
def decide(self, context: AgentContext) -> Optional[AgentAction]:
"""Use LLM to decide on action."""
self._set_state(AgentState.THINKING)
try:
from config import config
from tools import registry
tool_schemas = [
registry.get(name).to_llm_schema()
for name in self.tools
if registry.get(name)
]
messages = self._build_messages(context)
adapter = self._get_adapter()
llm_response = adapter.generate(
messages=messages,
tools=tool_schemas if tool_schemas else None,
temperature=self.llm_config.get("temperature", config.llm.temperature),
max_tokens=self.llm_config.get("max_tokens", config.llm.max_tokens)
)
if llm_response.tool_calls:
tool_call = llm_response.tool_calls[0]
return AgentAction(
tool_name=tool_call.tool_name,
arguments=tool_call.arguments,
reasoning=tool_call.reasoning
)
return None
except Exception as e:
logger.error(f"Agent {self.agent_id} decision failed: {e}")
self._set_state(AgentState.ERROR)
return None
def execute(self, action: AgentAction, context: AgentContext) -> AgentResult:
"""Execute tool action."""
self._set_state(AgentState.EXECUTING)
import time
start = time.perf_counter()
try:
from tools import invoke_tool
result = invoke_tool(action.tool_name, **action.arguments)
elapsed = (time.perf_counter() - start) * 1000
context.add_to_history(action.tool_name, result)
self._set_state(AgentState.COMPLETED)
return AgentResult(
success=result.get("success", False),
data=result,
message=f"Executed {action.tool_name}",
actions_taken=[action.tool_name],
execution_time_ms=elapsed
)
except Exception as e:
logger.error(f"Agent {self.agent_id} execution failed: {e}")
self._set_state(AgentState.ERROR)
return AgentResult(
success=False,
data=None,
message=str(e)
)
class RuleBasedAgent(BaseAgent):
"""
Agent that uses predefined rules for decision making.
Useful for deterministic behavior in Guided mode.
"""
def __init__(self,
agent_id: str,
role: AgentRole,
rules: List[Callable[[AgentContext], Optional[AgentAction]]],
tools: List[str] = None):
super().__init__(agent_id, role, tools)
self.rules = rules
def decide(self, context: AgentContext) -> Optional[AgentAction]:
"""Apply rules to decide action."""
self._set_state(AgentState.THINKING)
for rule in self.rules:
action = rule(context)
if action is not None:
return action
return None
def execute(self, action: AgentAction, context: AgentContext) -> AgentResult:
"""Execute action using tools."""
self._set_state(AgentState.EXECUTING)
import time
start = time.perf_counter()
try:
from tools import invoke_tool
result = invoke_tool(action.tool_name, **action.arguments)
elapsed = (time.perf_counter() - start) * 1000
context.add_to_history(action.tool_name, result)
self._set_state(AgentState.COMPLETED)
return AgentResult(
success=result.get("success", False),
data=result,
message=f"Executed {action.tool_name}",
actions_taken=[action.tool_name],
execution_time_ms=elapsed
)
except Exception as e:
self._set_state(AgentState.ERROR)
return AgentResult(
success=False,
data=None,
message=str(e)
)
|