Agent Team Design
SkillDocs & knowledgeYour AI can design a complete multi-agent team for you, with clear role boundaries, a handoff flow, and a visible agents folder. The skill also covers when to bring in named roles like PM Soul, Memory Curator, Policy Gate, and an evaluation role. It works even if all you ask for is a meta-agent or an agent operating system.
Available today. Use it from your connected AI after setup.
No other account needed.
Tell your AI what kind of team you want, even a single phrase like 'I want a meta-agent'. It will walk through the design and set up the agent-team repo.
Then ask your AI: use the Agent Team Design skill
What your AI can do with it
- Design a new multi-agent team from a simple request
- Set up a visible agents folder for the team
- Define role boundaries so each agent knows its job
- Plan the handoff flow for passing work between agents
- Add named roles such as PM Soul, Memory Curator, Policy Gate, and an evaluation role
- Create an agent-team repo even when you only ask for a meta-agent
What this skill tells your AI
The instructions your AI receives, as published by agentlas-ai/agentlas-os in skills/agent-team-design/SKILL.md and read by ahel’s review.
Procedure
- Define the user job and target runtime.
- Start with one agent. Add team roles only when routing, memory, review, policy, or parallel ownership is useful.
- Use a visible public tree:
agents/10-single-agent-builder/agent.mdagents/20-multi-agent-team-builder/agent.mdagents/30-agentlas-packager/agent.mdagents/40-session-agent-builder/agent.md
- Add
skills/<capability>/SKILL.mdfor reusable procedures. - Add
.agentlas/company-blueprint.jsonfor topology.
Output
Return the role list, handoff direction, memory owner, policy owner, and eval owner.
Signals
- GitHub stars
- 1k
- Forks
- 103
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
agent-team-design- Source
- github.com/agentlas-ai/agentlas-os