AgenticAI Framework

The Enterprise-Grade Python Framework for Building Production AI Agent Systems

Get Started Star on GitHub

Framework at a Glance¶

400+

Total Modules

237

Enterprise Modules

35+

Built-in Tools

12

Evaluation Tiers

6

Communication Protocols

14

Enterprise Categories

Why AgenticAI Framework?¶

Advanced Memory System

7 specialized memory managers with semantic search, compression, and persistence. Never lose context again.

Explore Memory
Multi-Agent Orchestration

Build complex AI teams with hierarchical workflows, parallel execution, and intelligent routing.

Learn Orchestration
6 Communication Protocols

HTTP, WebSocket, SSE, MQTT, gRPC, and STDIO - connect agents any way you need.

View Protocols
35+ Production Tools

Search, code execution, file operations, database queries, web scraping, and more built-in.

Browse Tools
Enterprise Security

Secrets management, input validation, output sanitization, RBAC, and compliance auditing.

Security Guide
12-Tier Evaluation

Comprehensive testing framework for model quality, security, cost, and business metrics.

Evaluation System

Architecture Overview¶

Text Only

┌─────────────────────────────────────────────────────────────────────────────┐
│ AgenticAI Framework Architecture │
├─────────────────────────────────────────────────────────────────────────────┤
│ │
│ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │
│ │ Agent 1 │ │ Agent 2 │ │ Agent 3 │ │ Agent N │ │
│ │ (Leader) │ │ (Researcher)│ │ (Writer) │ │ (Specialist)│ │
│ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ │
│ │ │ │ │ │
│ └────────────────┼────────────────┼────────────────┘ │
│ │ │ │
│ ┌───────────────────────┴────────────────┴───────────────────────┐ │
│ │ Orchestration Layer │ │
│ │ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │ │
│ │ │ Workflow │ │ Team │ │ Router │ │ │
│ │ │ Manager │ │ Manager │ │ Engine │ │ │
│ │ └─────────────┘ └─────────────┘ └─────────────┘ │ │
│ └─────────────────────────────────────────────────────────────────┘ │
│ │ │
│ ┌─────────────────────────────────┴─────────────────────────────────┐ │
│ │ Core Services Layer │ │
│ ├────────────┬────────────┬────────────┬────────────┬───────────────┤ │
│ │ Memory │ State │ Knowledge │ Tool │ LLM │ │
│ │ Managers │ Managers │ Base │ Registry │ Providers │ │
│ │ (7 types) │ (7 types) │ │ (35+) │ (Multiple) │ │
│ └────────────┴────────────┴────────────┴────────────┴───────────────┘ │
│ │ │
│ ┌─────────────────────────────────┴─────────────────────────────────┐ │
│ │ Infrastructure Layer │ │
│ ├──────────────┬──────────────┬──────────────┬──────────────────────┤ │
│ │ Communication│ Tracing & │ Security & │ Guardrails & │ │
│ │ (6 protocols)│ Monitoring │ Compliance │ Validation │ │
│ └──────────────┴──────────────┴──────────────┴──────────────────────┘ │
│ │
└─────────────────────────────────────────────────────────────────────────────┘

Quick Start¶

Installation¶

Bash
1	`pip install agenticaiframework`

Create Your First Agent¶

Python
from agenticaiframework import Agent, MemoryManager
import logging

logger = logging.getLogger(__name__)

# Initialize memory management
memory = MemoryManager()

# Create and run the agent
agent = Agent(
    name="research_assistant",
    role="Research Analyst",
    capabilities=["research", "analysis"],
)
agent.start()
result = agent.act("Research the latest trends in AI agents")
logger.info("Result: %s", result)

Build a Multi-Agent Team¶

Python
from agenticaiframework import Agent, AgentManager

# Create specialized agents
researcher = Agent(
    name="researcher", role="Research Expert",
    capabilities=["web", "text"],
)
writer = Agent(
    name="writer", role="Content Writer",
    capabilities=["text"],
)
editor = Agent(
    name="editor", role="Quality Editor",
    capabilities=["text"],
)

# Register with manager
manager = AgentManager()
for a in (researcher, writer, editor):
    manager.register_agent(a)

# Start agents
for a in manager.list_agents():
    a.start()

See Full Quick Start Guide

Core Modules¶

Agents

Create intelligent agents with customizable roles, goals, tools, and behaviors.

Python
from agenticaiframework import Agent, AgentConfig

agent = Agent(config=AgentConfig(
    name="assistant",
    role="AI Assistant",
    capabilities=["search", "calculator"],
))

Learn More

Memory

7 specialized memory managers for different use cases.

Manager	Purpose
`MemoryManager`	General-purpose memory
`AgentMemoryManager`	Agent-specific memories
`WorkflowMemoryManager`	Workflow state tracking
`OrchestrationMemoryManager`	Multi-agent coordination
`KnowledgeMemoryManager`	Knowledge base storage
`ToolMemoryManager`	Tool execution history
`SpeechMemoryManager`	Voice interaction data

Explore Memory

State Management

7 dedicated state managers for complete system control.
- AgentStateManager - Agent lifecycle
- WorkflowStateManager - Workflow execution
- ConversationStateManager - Conversation context
- TaskStateManager - Task tracking
- ContextStateManager - Context windows
- ToolStateManager - Tool states
- MemoryStateManager - Memory states
State Guide
:satellite_antenna:{ .lg } Communication

6 protocols for any integration scenario.
- HTTP - REST API communication
- WebSocket - Real-time bidirectional
- SSE - Server-sent events streaming
- MQTT - IoT message queuing
- gRPC - High-performance RPC
- STDIO - Process communication
Protocols

35+ Built-in Tools¶

Search & InformationCode & DevelopmentFile & DataDatabase & StorageAI & MLUtilities

Web Search - Multi-engine search
News Search - Real-time news
Wikipedia - Encyclopedia access
URL Fetch - Web scraping
DNS Lookup - Network tools

Python REPL - Code execution
Package Manager - Dependency management
Test Runner - Automated testing
Code Analysis - Static analysis
Debugger - Issue diagnosis

File Operations - CRUD operations
CSV Handler - Tabular data
JSON Tools - JSON manipulation
Data Analysis - Statistical analysis
Text Search - Content search

PostgreSQL - SQL databases
MongoDB - NoSQL storage
Redis - Cache operations
Vector Store - Embeddings
Cloud Storage - S3, GCS, Azure

Embeddings - Vector generation
Image Generation - DALL-E, Stable Diffusion
Speech-to-Text - Audio transcription
Text-to-Speech - Voice synthesis
Vision - Image analysis

DateTime - Time operations
Encryption - Security tools
Email - SMTP/IMAP
Notifications - Alerts & webhooks
Clipboard - System clipboard

Complete Tool Reference

12-Tier Evaluation System¶

Comprehensive evaluation framework for production AI systems:

Tier	Category	Evaluates
1	Model Quality	Accuracy, coherence, hallucination detection
2	Task Performance	Completion rate, efficiency, error handling
3	Tool Effectiveness	Tool selection, execution success, latency
4	Memory & RAG	Retrieval accuracy, context relevance
5	Autonomy	Decision quality, self-correction
6	Security	Prompt injection, data leakage, PII handling
7	Cost Optimization	Token usage, API costs, resource efficiency
8	Human Alignment	Feedback incorporation, preference matching
9	Drift Detection	Performance degradation, distribution shift
10	A/B Testing	Variant comparison, statistical significance
11	Canary Deployment	Gradual rollout, risk mitigation
12	Workflow Analytics	End-to-end metrics, bottleneck detection

Full Evaluation Guide

Enterprise Features¶

Security & Compliance
- Secrets management with encryption
- Input validation & sanitization
- RBAC with fine-grained permissions
- Audit logging & compliance trails
- PII detection & masking
- SOC 2, HIPAA, GDPR ready
Monitoring & Observability
- OpenTelemetry integration
- Distributed tracing
- Custom metrics & dashboards
- Real-time alerting
- Performance profiling
- Error tracking & analysis
Scalability & Performance
- Async execution support
- Connection pooling
- Caching layers
- Load balancing ready
- Rate limiting
- Circuit breakers
DevOps & Deployment
- Docker & Kubernetes ready
- CI/CD pipeline templates
- Infrastructure as Code
- Multi-environment configs
- Blue-green deployments
- Rollback mechanisms

Speech Processing¶

Full-featured speech-to-text and text-to-speech capabilities:

Python
from agenticaiframework import SpeechMemoryManager, AgentConfig

# Initialize speech-enabled agent
speech_memory = SpeechMemoryManager()
agent = Agent(
    config=AgentConfig(
        name="voice_assistant",
        speech_enabled=True
    ),
    speech_memory=speech_memory
)

# Process voice input
transcript = agent.transcribe_audio("user_audio.wav")
response = agent.execute(transcript)
audio = agent.synthesize_speech(response.output)

Supported Providers:

OpenAI Whisper (STT)
Google Cloud Speech (STT/TTS)
Azure Cognitive Services (STT/TTS)
Amazon Transcribe/Polly (STT/TTS)
ElevenLabs (TTS)

Speech Documentation

Documentation¶

Community & Support¶

:material-github: GitHub :material-bug: Issues :material-forum: Discussions :material-heart: Contributing

License¶

AgenticAI Framework is released under the MIT License.

Text Only

MIT License

Copyright (c) 2024 Sathish Babu

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software...

Full License

Built with for the AI Agent Community

Get Started Now