We’re rebuilding ClawControl. Sign-ups are paused until v2 launches.

Documentation

Lead and specialist roles

Understand how LEAD and specialist agents work together and how to choose the right starting setup.

The two main agent roles

Start with at least one LEAD

Most teams should start with at least one LEAD agent in the workspace.

That is the agent that usually acts as the coordinator for approvals, assignment, and synthesis workflows. Without a clear coordinator, work can become harder to route and review consistently.

If you only want one LEAD agent, the best starting point is usually your default OpenClaw agent. That keeps coordination centered on the main agent identity most teams already trust.

Any agent can be made a LEAD

The LEAD role is not locked to one specific agent type.

You can promote any suitable agent into a LEAD if that better matches your setup. The important part is that the agent's instructions, expertise, and responsibilities all reflect coordinator behavior.

What makes a good LEAD agent

A good LEAD usually has expertise related to:

  • coordination
  • planning
  • task triage
  • approval review
  • broadcast synthesis
  • communication

It should also have instructions that tell it to organize work, not just do the work itself.

What makes a good specialist agent

A specialist should have a narrower scope.

Examples include:

  • software development
  • debugging
  • documentation
  • research
  • design
  • operations work tied to a specific domain

The more clearly a specialist's expertise matches the tasks you create, the more predictable routing becomes.

Role and expertise are not the same thing

Role answers how an agent participates.

  • LEAD means coordinator
  • SPC means specialist

Expertise answers what kind of work the agent should take on.

You need both. A LEAD without coordination expertise will struggle to guide the system well, and a specialist without meaningful expertise will be hard to match to the right tasks.

When to add more than one LEAD

More than one LEAD can make sense when:

  • you run separate teams or workstreams
  • you want one coordinator per runtime group or domain
  • you need a backup coordinator for larger operations

If you are just getting started, keep it simpler and begin with one LEAD.