#Proposal: Ephemeral container based deployment for Openclaw

1 messages · Page 1 of 1 (latest)

bright yarrow
#

Current docs focus on using Openclaw deployment as pet. However, all the capabilities exist in the docs. Proposal to create a dedicated chapter for deploying Openclaw on ephemeral container platforms.

#

Some more details on the approach.

What gets backed up to object storage (can't be regenerated):

  • credentials/ - auth tokens from interactive flows (WhatsApp QR sessions, etc.)
  • workspace/MEMORY.md + workspace/memory/ - agent memory
  • agents/<id>/sessions/ - conversation history and transcripts

What gets regenerated every boot (no backup needed):

  • openclaw.json - from templates + env vars
  • Channel config - tokens as env vars
  • Provider config - API keys as env vars
  • Agent identity files - fetched from object storage, or OpenClaw defaults

Questions:

  1. Does this align with how you see container deployments?
  2. Anything missing from the backup strategy?
  3. Any concerns with generating openclaw.json externally rather than using the CLI?