Choose theme
AI Resources
AG2
AG2 v1 is a Python framework for building agents that can use models and tools, pause for human input, and coordinate through a multi-agent Network.
The current package uses an Agent class and a protocol-driven Network with a central hub and typed channels. The older AutoGen-derived classes and patterns now live in the separately maintained AG2 Classic repository. Use this as a first read, not a recommendation. Open the original project before trusting details like terms, limits, privacy, cost, setup, or safety.
What it is
Agents connected through a Network
A single Agent can call models and Python tools. When several agents need to work together, AG2 can connect them through a hub with conversation, consulting, discussion, or workflow channels.
Why it stands out
A clear split between v1 and Classic
AG2 v1 is not a drop-in upgrade from AG2 Classic. Existing projects using ConversableAgent, GroupChat, swarms, or the autogen import should follow the maintained Classic package and its migration guidance.
Availability
Repository, docs, examples, and PyPI
AG2 v1 is available on PyPI, with a GitHub repository, current documentation, runnable examples, and a Network guide for its hub-and-channel orchestration.
Why it matters
What makes it useful
If a Python task needs a model to call a function and use its result, AG2 lets you attach that function as a tool to one Agent. A Network is optional: add it when several named agents need to exchange work with recorded messages and a defined turn order.
What to know
Where it fits
Start with a model-provider configuration, a Python tool, and a task for one Agent. For a planner and reviewer working together, the Network supplies the participants and channel rules; your application still defines the task and checks the result.
Notable points
What stands out
Check the import before following an example: import ag2 is the current v1 framework, while import autogen and classes such as ConversableAgent or GroupChat belong to AG2 Classic.
Before using
What to review
Whether the project is starting on AG2 v1 or maintaining AG2 Classic, because their agent models, orchestration, imports, and documentation differ.
The model-provider setup, API-key handling, tool permissions, and human-input points in any workflow.
Whether a Network is actually needed, or whether one Agent with a small set of tools would be easier to inspect.
The installed package version against the security advisories. AG2 lists 1.1.0 as the patch for several filesystem, skill-installation, and approval-context issues; the matching Classic entries list no patched version. Check the package and affected feature separately. Later LocalShell and served-over-wire NetworkHub advisories name AG2 1.1.2 as patched; the Hub advisory excludes in-process hubs without wire transport and lists no patch for its autogen mirror. Match your installed version, package and used feature to each current advisory.
Reader fit
Who may find it relevant
Developers and researchers comparing multi-agent channel patterns and orchestration styles.
Readers studying how current AG2 Agents use tools, ask for human input, and coordinate through a Network hub and typed channels.
Not the first stop for readers who want a visual workflow builder, a voice-agent transport stack, or a simple consumer app.
Editorial note
Why LifeHubber lists it
AG2 is useful when a planner needs a bounded answer from a specialist agent before continuing its own task. Its consulting protocol gives that delegation a defined end after one exchange, rather than making every collaboration an ongoing conversation. For one agent asking another a single question, choose a consulting channel: it allows one question and one reply, then closes automatically. Choose conversation when the two agents need an ongoing exchange, and set its explicit close or time-to-live rule. The channel choice determines how the exchange ends; naming a second agent does not define that stopping point.
Source links
Source materials
Reader note
Before relying on this entry
LifeHubber lists entries to help readers inspect AI projects, not to endorse them or prove they are safe, suitable, accurate, maintained, or right for a specific use. We do not verify every entry in depth. Before relying on anything listed, review the original materials, terms, privacy practices, limits, and risks that matter for your situation.
What to explore next
Compare how agents coordinate work.
AG2 v1 coordinates agents through a hub and typed channels, with tools and human input available inside the work. Compare a code-first toolkit, a Python framework that separates agent teams from explicit flows, or the wider framework landscape.
More in AI Agents
Keep browsing this category
Explore more AI agent projects.
Paperclip
paperclipai/paperclip
A self-hosted server and dashboard for coordinating agent teams through companies, goals, roles, issues, heartbeats, budgets, approvals, and persistent activity records.
AgentMemory
rohitg00/agentmemory
A self-hosted coding-agent memory server with session capture, keyword recall, optional local or remote embeddings, a viewer, and integrations through hooks, MCP, and REST.
Memvid
memvid/memvid
A Rust-based memory layer for AI agents that keeps documents, embeddings, search indexes, metadata, and recovery state in one portable .mv2 file, with CLI, Python, Node.js, and framework integration paths.
For project maintainers
Listed here? You can use the badge.
If you maintain a project with a current LifeHubber listing, you may add the optional “Listed on LifeHubber AI Resources” badge to its README, docs, or website. No introduction or permission request is needed.
