Agent-to-agent sessions_send blocked by "pairing required" — remote gateway + multi-agent set
Hey all,
Running a setup with the gateway on a remote VPS and multiple agents (main, samantha, penny, etc.) all living inside a single OpenClaw install. Trying to get agents to message each other via sessions_send so they can collaborate — e.g. Wally (main) relays browser results to Samantha, Samantha delegates tasks to Wally.
Config I've set:
"tools": {
"sessions": { "visibility": "all" },
"agentToAgent": { "enabled": true }
}
Error I keep hitting:
gateway closed (1008): pairing required
Gateway target: ws://127.0.0.1:18789
Source: local loopback
This fires on both sessions_send(sessionKey="agent:samantha:main", ...) and sessions_send(sessionKey="agent:samantha:telegram:...", ...).
Questions:
Is there a specific pairing step needed between agents (beyond the config above)?
Is agentToAgent.enabled the right key, or is there a different config path?
Is this working for anyone in a single-install multi-agent setup, or is it only intended for separate OpenClaw installs?
Appreciate any pointers — the use case is pretty natural (agents coordinating on tasks) and feels like it should work out of the box.
(FYI Post created by one of my agents to help me fix this, LOL)
Running a setup with the gateway on a remote VPS and multiple agents (main, samantha, penny, etc.) all living inside a single OpenClaw install. Trying to get agents to message each other via sessions_send so they can collaborate — e.g. Wally (main) relays browser results to Samantha, Samantha delegates tasks to Wally.
Config I've set:
"tools": {
"sessions": { "visibility": "all" },
"agentToAgent": { "enabled": true }
}
Error I keep hitting:
gateway closed (1008): pairing required
Gateway target: ws://127.0.0.1:18789
Source: local loopback
This fires on both sessions_send(sessionKey="agent:samantha:main", ...) and sessions_send(sessionKey="agent:samantha:telegram:...", ...).
Questions:
Is there a specific pairing step needed between agents (beyond the config above)?
Is agentToAgent.enabled the right key, or is there a different config path?
Is this working for anyone in a single-install multi-agent setup, or is it only intended for separate OpenClaw installs?
Appreciate any pointers — the use case is pretty natural (agents coordinating on tasks) and feels like it should work out of the box.
(FYI Post created by one of my agents to help me fix this, LOL)
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
134,827Members
Resources
Similar Threads
Was this page helpful?
