feat: add autogen — expert skill for conversational multi-agent AI #82

Closed
opened 2026-07-09 14:41:21 -04:00 by jasper · 1 comment
Contributor

Summary

Add an autogen expert skill for conversational multi-agent AI. AutoGen by Microsoft Research provides a distinguished approach to multi-agent systems: agents communicate through structured conversations with cancellable, delegatable tasks.

Why This Skill

AutoGen's conversational agent model is categorically distinct from LangGraph (state-machine graphs), CrewAI (role-based crews), and PydanticAI (type-safe individual agents). It fills the conversation-driven multi-agent niche in the portfolio.

Key Differentiators

AutoGen uses agent conversations as the orchestration primitive, with GroupChat for speaker routing, built-in code executors, cancellation tokens for agent tasks, and nested chat chains for delegation.

Scope

  • Deep research: AutoGen architecture, nested chats, cancellation tokens, code executors
  • Vault notes: atoms + molecules on AutoGen conversational model
  • Expert skill: SKILL.md + references + templates + scripts
  • Greenfield SkillOpt: 3 epochs
  • PR: Single PR to magnus/agent-skills

Key References

Acceptance Criteria

  • Deep research pyramid
  • Vault clipping + 20+ atoms + 4 molecules
  • Agent skill with 7+ references, 3+ templates, 1+ script
  • Greenfield SkillOpt 3 epochs
  • PR merged to main
## Summary Add an autogen expert skill for conversational multi-agent AI. AutoGen by Microsoft Research provides a distinguished approach to multi-agent systems: agents communicate through structured conversations with cancellable, delegatable tasks. ## Why This Skill AutoGen's conversational agent model is categorically distinct from LangGraph (state-machine graphs), CrewAI (role-based crews), and PydanticAI (type-safe individual agents). It fills the conversation-driven multi-agent niche in the portfolio. ## Key Differentiators AutoGen uses agent conversations as the orchestration primitive, with GroupChat for speaker routing, built-in code executors, cancellation tokens for agent tasks, and nested chat chains for delegation. ## Scope - Deep research: AutoGen architecture, nested chats, cancellation tokens, code executors - Vault notes: atoms + molecules on AutoGen conversational model - Expert skill: SKILL.md + references + templates + scripts - Greenfield SkillOpt: 3 epochs - PR: Single PR to magnus/agent-skills ## Key References - https://microsoft.github.io/autogen - https://github.com/microsoft/autogen ## Acceptance Criteria - Deep research pyramid - Vault clipping + 20+ atoms + 4 molecules - Agent skill with 7+ references, 3+ templates, 1+ script - Greenfield SkillOpt 3 epochs - PR merged to main
Author
Contributor

Resolved by merged PR #86: #86

Filed by Jasper (AI agent on behalf of Magnus Hedemark)

Resolved by merged PR #86: https://git.brandyapple.com/magnus/agent-skills/pulls/86 Filed by Jasper (AI agent on behalf of Magnus Hedemark)
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
magnus/agent-skills#82
No description provided.