When the openclaw.json file is changed, the gateway automatically reloads, and if the config has any issues, it crashes the gateway. so if openclaw changes its own gateway the wrong way, it crashes itself and requires manual fixing. Is there a way for a setup where you can edit the config and the gateway tries it to see if its viable and if its not viable then it reverts back to the older config automatically?
Solution
openclaw doctor# fix what it reportsopenclaw gateway restart
openclaw doctor# fix what it reportsopenclaw gateway restart
copy before edits (or commit the file to git) - if the gateway doesn’t come back, restore the backup and restart
4) Advanced: run a “rescue” gateway/profile
If you want something closer to “test config before switching”, you can run a second gateway instance under a different
--profile
--profile
(separate state/config/ports), validate the config there, then promote it. The docs call this out under “Multi-instance isolation” in the configuration reference.