API Reference

Complete API documentation for SPADE_LLM components.

Core Components

• Agent - LLMAgent and ChatAgent classes
• Behaviour - LLMBehaviour implementation
• Providers - LLM provider interfaces
• Tools - Tool system and LLMTool class
• Memory - Memory system API for agent learning and persistence
• Human Interface - Human-in-the-loop API and integration
• Guardrails - Content filtering and safety controls
• Context - Context and conversation management
• Routing - Message routing system

Quick Reference

Creating Agents

from spade_llm import LLMAgent, LLMProvider

provider = LLMProvider.create_openai(api_key="key", model="gpt-4o-mini")
agent = LLMAgent(jid="agent@server.com", password="pass", provider=provider)

Creating Tools

from spade_llm import LLMTool

async def my_function(param: str) -> str:
    return f"Result: {param}"

tool = LLMTool(
    name="my_function",
    description="Description of function",
    parameters={"type": "object", "properties": {"param": {"type": "string"}}, "required": ["param"]},
    func=my_function
)

Message Routing

def router(msg, response, context):
    if "technical" in response.lower():
        return "tech@example.com"
    return str(msg.sender)

agent = LLMAgent(..., routing_function=router)

Examples

See Examples for complete working code examples.

Type Definitions

Common Types

# Message context
ContextMessage = Union[SystemMessage, UserMessage, AssistantMessage, ToolResultMessage]

# Routing result
RoutingResult = Union[str, List[str], RoutingResponse, None]

# Tool parameters
ToolParameters = Dict[str, Any]  # JSON Schema format

Error Handling

All SPADE_LLM components use standard Python exceptions:

• ValueError - Invalid parameters or configuration
• ConnectionError - Network or provider connection issues
• TimeoutError - Operations that exceed timeout limits
• RuntimeError - General runtime errors

Configuration

Environment Variables

OPENAI_API_KEY=your-api-key
OLLAMA_BASE_URL=http://localhost:11434/v1
LM_STUDIO_BASE_URL=http://localhost:1234/v1

Provider Configuration

# OpenAI
provider = LLMProvider.create_openai(api_key="key", model="gpt-4o-mini")

# Ollama  
provider = LLMProvider.create_ollama(model="llama3.1:8b")

# LM Studio
provider = LLMProvider.create_lm_studio(model="local-model")

For detailed API documentation, see the individual component pages.