AG2

In this catalog it is the Frameworks option when you need to compose multi-step agent loops and tool handoffs. Keep AG2 on the observe/propose side of the loop: useful signals and drafts, with human or policy authority still required before anything irreversible runs.

AG2 runs conversational multi-agent chats: agents exchange messages, call tools, and terminate when a speaker decides the task is done.

1. Define agents (system message, LLM, tools) and a chat/group pattern.
2. Start a conversation with a seed task; the runtime routes speaker turns.
3. Tool calls execute in the loop; results return as messages.
4. Persist or hand off the transcript when the chat ends.

Related: autogen · crew-ai · topics/12-central-nervous-system

• Use when: AG2 is the community-governed continuation of the original AutoGen ecosystem for conversational multi-agent systems — preserving and evolving the multi-agent chat patterns outside a single-vendor path matches the bottleneck and you specifically need to compose multi-step agent loops and tool handoffs.
• Skip when: another pack already covers the need with less ops surface, or AG2 would become a second place people retype truth.
• Prefer instead: compare Autogen, Crew AI before locking the cell.

• Ops: confirm license, hosting, and release cadence for AG2 (ag2ai/ag2) before relying on it in a live corridor.
• Integration: bad credentials, scopes, or schema drift produce empty retrievals or runaway tool/model spend without an obvious UI alarm.
• Scope: Frameworks components support agents; they are not the company ledger or the approval authority.