Core Modules¶

SikkaAgent is built on a modular architecture where each component serves a specific purpose in the agent framework.

Overview¶

The core modules of SikkaAgent work together to create a flexible, powerful platform for building AI agent systems:

Module	Description	Key Features
Agents	Agent implementations with different capabilities	Conversation, task execution, knowledge integration
Models	Interface to various AI models	Unified model API, provider abstraction
Memories	Contextual memory systems	Conversation history, prioritization strategies
Tools	Specialized capabilities for agents	Web browsing, search, file manipulation, media analysis
Retrievers	Information retrieval components	Vector search, semantic querying
Storages	Data persistence solutions	Memory storage, vector databases
Workflows	Multi-agent orchestration	DAG-based pipelines, role-playing simulations
Prompts	Prompt management and templating	System prompts, dynamic templates
Knowledge	Knowledge Base	Create Knowledge, Manage Knowledge Base, Query Knowledge Base
Documents	Document Processing	Document Processing, Handling

Module Relationships¶

The core modules interact in a layered architecture:

┌─────────────────────────────────────┐
│              Agents                 │
└───────────────┬─────────────────────┘
                │
┌───────────────┼─────────────────────┐
│   Models      │       Workflows     │
└───────────────┼─────────────────────┘
                │
┌───────────────┼─────────────────────┐
│  Memories     │     Retrievers      │
└───────────────┼─────────────────────┘
                │
┌───────────────┼─────────────────────┐
│   Tools       │      Storages       │
└───────────────┼─────────────────────┘
                │
┌───────────────┼─────────────────────┐
│   Knowledge   │      Documents      │
└───────────────┼─────────────────────┘
                │
┌───────────────┴─────────────────────┐
│             Prompts                 │
└─────────────────────────────────────┘

Agents combine various modules to provide complete conversational or task-focused interfaces
Models power the reasoning capabilities of agents through LLM integration
Memories maintain context across interactions using various storage backends
Tools extend agent capabilities with specialized functions
Retrievers enable intelligent information access from various sources
Storages handle data persistence and retrieval for various components
Workflows orchestrate collaboration between multiple agents
Prompts underpin everything by structuring how agents communicate with models

Integration Patterns¶

SikkaAgent modules are designed to be used together in various combinations:

Agent + Memory + Tools¶

from sikkaagent.agents import ChatAgent
from sikkaagent.memories import Memory
from sikkaagent.storages import InMemoryStorage
from sikkaagent.tools import SearchToolkit
from sikkaagent.models import ModelConfigure
from sikkaagent.utils.enums import ModelPlatformType

# Initialize model
model = ModelConfigure(
    model="llama3.1:8b",
    model_platform=ModelPlatformType.OLLAMA,
    url="http://localhost:11434/v1"
)

# Create agent with memory and tools
agent = ChatAgent(
    model=model,
    memory=InMemoryStorage(),
    tools=[*SearchToolkit().get_tools()]
)

# Interact with the agent
response = agent.step("What were the major AI developments in 2023?")

Retriever + Knowledge + Agent¶

from sikkaagent.agents import ChatAgent
from sikkaagent.knowledge import AgentKnowledge
from sikkaagent.retrievers.sentence_transformers_embeddings import SentenceTransformerEmbeddings
from sikkaagent.storages.qdrant import Qdrant

# Create embedder and storage
embedder = SentenceTransformerEmbeddings()
vector_db = Qdrant(collection="knowledge_base", embedder=embedder)

# Create knowledge with vector database
knowledge = AgentKnowledge(vector_db=vector_db)

# Add to agent
agent = ChatAgent(
    model=model,
    knowledge=knowledge,
    add_references=True
)

Workflow + Multiple Agents¶

from sikkaagent.workflows.dag import StateGraph, State, START, END
from sikkaagent.agents import ChatAgent

# Create specialized agents
research_agent = ChatAgent(model=model, system_prompt="You research facts...")
writing_agent = ChatAgent(model=model, system_prompt="You write content...")

# Create workflow
graph = StateGraph(State)
graph.add_node("research", research_agent)
graph.add_node("writing", writing_agent)
graph.add_edge(START, "research")
graph.add_edge("research", "writing")
graph.add_edge("writing", END)

# Run workflow
workflow = graph.compile()
result = workflow.invoke({"query": "Write about quantum computing"})