Fusion 360 MCP

""" LLM Response Schema Unified response structure from all LLM providers """ from typing import Any, Dict, List, Optional, Literal from pydantic import BaseModel, Field from datetime import datetime from .fusion_action import FusionAction, ActionSequence class LLMResponseMetadata(BaseModel): """Metadata about the LLM response""" provider: Literal["ollama", "openai", "gemini", "claude"] = Field(description="Provider name") model: str = Field(description="Model name") timestamp: datetime = Field(default_factory=datetime.now, description="Response timestamp") tokens_used: Optional[int] = Field(default=None, description="Tokens consumed") latency_ms: Optional[float] = Field(default=None, description="Response latency") temperature: Optional[float] = Field(default=None, description="Temperature used") class LLMError(BaseModel): """Error structure for LLM failures""" type: str = Field(description="Error type (timeout, api_error, validation_error, etc.)") message: str = Field(description="Error message") provider: str = Field(description="Provider that failed") retry_count: int = Field(default=0, description="Number of retries attempted") recoverable: bool = Field(default=True, description="Whether error is recoverable") class ClarifyingQuestion(BaseModel): """Structure for questions when user intent is ambiguous""" question: str = Field(description="The clarifying question") context: str = Field(description="Why this question is needed") suggestions: List[str] = Field(default_factory=list, description="Suggested answers") class LLMResponse(BaseModel): """Unified LLM response structure""" success: bool = Field(description="Whether the response was successful") provider: str = Field(description="Provider name") model: str = Field(description="Model name") raw_output: str = Field(description="Raw text output from model") action: Optional[FusionAction] = Field(default=None, description="Parsed Fusion action") action_sequence: Optional[ActionSequence] = Field(default=None, description="Multiple actions") clarifying_questions: List[ClarifyingQuestion] = Field( default_factory=list, description="Questions for ambiguous requests" ) reasoning: Optional[str] = Field(default=None, description="Model's reasoning process") metadata: LLMResponseMetadata = Field(description="Response metadata") error: Optional[LLMError] = Field(default=None, description="Error if failed") class Config: json_schema_extra = { "example": { "success": True, "provider": "openai", "model": "gpt-4o-mini", "raw_output": "I'll create a 20mm cube...", "action": { "action": "create_box", "params": {"width": 20, "height": 20, "depth": 20, "unit": "mm"}, "explanation": "Creating a 20mm cube as requested" }, "metadata": { "provider": "openai", "model": "gpt-4o-mini", "tokens_used": 150, "latency_ms": 1200 } } } class MCPResponse(BaseModel): """Response sent back to Fusion 360""" status: Literal["success", "error", "clarification_needed", "partial"] = Field( description="Response status" ) llm_response: Optional[LLMResponse] = Field(default=None, description="LLM response data") message: str = Field(description="Human-readable status message") actions_to_execute: List[FusionAction] = Field( default_factory=list, description="Actions ready for execution" ) warnings: List[str] = Field(default_factory=list, description="Warning messages") class Config: json_schema_extra = { "example": { "status": "success", "message": "Action generated successfully", "actions_to_execute": [ { "action": "create_box", "params": {"width": 20, "height": 20, "depth": 20}, "explanation": "Creating cube" } ], "llm_response": { "success": True, "provider": "openai", "model": "gpt-4o-mini" } } }