Let your pods work together
Ask Claude in one project to hand work to another of your pods and bring the result back.
Separate pods are useful when projects need different tools, access, or context. They do not have to be isolated from one another: Claude in one pod can ask Claude in another of your pods to do a task and report back.
When this helps
- A research pod prepares approved source material and asks a website pod to update a page.
- A product pod asks a documentation pod to check whether a shipped change needs new guidance.
- A coordinator pod asks several project pods for status and combines their answers.
- A monitoring pod reports a finding to the pod that owns the affected application.
Ask the docs pod to check whether the new account flow is documented. Have it make a local draft if needed, then bring me its summary. Do not publish anything.
You can address a pod by the name you gave it. Claude handles delivery and the reply; there is no inbox you need to operate during normal use.
What happens if the other pod is unavailable
If the receiving pod is running, the message wakes its agent with a new task. If it is suspended, the message waits and is delivered after that pod resumes. Delivery is durable, but work cannot run while the receiving pod is stopped.
The safety boundary
Pod-to-pod messages can only travel between pods owned by the same account. One of your pods cannot discover or contact another person's pod.
A message also does not inherit your authority. The receiving agent treats it as information and still asks for the approvals it would normally need before pushing code, publishing content, contacting someone, spending money, or changing an external account.
Give each pod a clear job
Name pods by the project or responsibility they own, and ask for a result that is local to that workspace. Clear ownership makes handoffs easier to understand and reduces duplicated work.