Security advisory — Malicious litellm versions 1.82.7 and 1.82.8 were removed from PyPI (potential API key exfiltration). Uninstall them, rotate exposed credentials, and upgrade to a safe release (e.g. 1.82.9+ per upstream). Run pip show litellm to verify. PyPI · README

FAQ

Frequently asked questions about AgenticX.

FAQ

General

What is AgenticX?

AgenticX is a unified, production-ready Python framework for building multi-agent AI applications. It provides everything from a single-agent execution engine to complex multi-agent orchestration, memory systems, tool integration, and a full Studio UI.

How is AgenticX different from LangChain or CrewAI?

AgenticX is designed for production from day one:

  • Unified: One framework covers agents, tools, memory, orchestration, protocols (A2A/MCP), observability, and security
  • Studio: A full web UI and Desktop app for managing agents, sessions, and group chats
  • Enterprise-ready: Security layer, sandbox execution, audit logging, and session isolation built in
  • Multi-agent native: Avatar system, group chat, Meta-Agent CEO pattern, and team management out of the box

Is AgenticX open source?

Yes. AgenticX is licensed under AGPL-3.0.

Installation

What Python version is required?

Python 3.10 or higher.

How do I install optional features?

bash
1pip install "agenticx[all]"    # Everything
2pip install "agenticx[vector]" # Vector store support
3pip install "agenticx[doc]"    # Document parsing

LLM Providers

Which LLM providers are supported?

15+ providers including OpenAI, Anthropic, Ollama (local), Gemini, Kimi/Moonshot, MiniMax, Ark/VolcEngine, Zhipu, Qianfan, and Bailian/Dashscope.

Can I use local models?

Yes. Use the OllamaProvider to connect to locally running models via Ollama:

python
1from agenticx.llms import OllamaProvider
2llm = OllamaProvider(model="llama3.2", base_url="http://localhost:11434")

Does MiniMax support image inputs?

No. All minimax-m2* models do not support image or audio inputs. The framework will warn you if you try to send images to these models.

Tools & MCP

What is MCP?

Model Context Protocol is an open standard for connecting AI agents to tools and data sources. AgenticX includes an MCP Hub that can connect to multiple MCP servers simultaneously.

Can I use tools from other frameworks?

Yes. AgenticX can import tools from LangChain, use OpenAPI specs to auto-generate toolsets, or call any HTTP endpoint via Remote Tools.

Memory

Where is memory stored?

By default, memory is stored in SQLite at ~/.agenticx/workspace/. You can configure Redis or PostgreSQL for production deployments.

Does memory persist across restarts?

Yes. All memory backends (SQLite, Redis, PostgreSQL) persist data between sessions.

Studio & Desktop

How do I start the Studio UI?

bash
1agx serve --port 8000
2# Open http://localhost:8000

What is the Desktop app?

The Machi Desktop app is an Electron-based application that wraps the Studio UI with native OS features like system tray, notifications, and multi-window support. It supports session restoration after restart.

Contributing

How can I contribute?

See CONTRIBUTING.md on GitHub.

How do I report a bug?

Open an issue on GitHub Issues.