[Q] how are upgrades handled?

[ralyodio]

openclaw gets upgraded almost daily, how is this handled?

[CorellisOrg]

Good question! Corellis is a configuration and governance layer that sits on top of OpenClaw — it doesn't patch or modify OpenClaw's source code. So the short answer: OpenClaw upgrades just work.

Here's how it breaks down:

What Corellis manages (all user-space files):

• AGENTS.md, SOUL.md, USER.md — agent behavior and governance
• MEMORY.md + memory/ — long-term memory and daily logs
• skills/ and company-skills/ — reusable task instructions
• Shared knowledge in /shared/ — cross-agent coordination

What OpenClaw manages (runtime):

• Tool execution, model routing, message handling, sessions
• OpenClaw upgrades via npm update -g openclaw — this only touches the runtime

Since Corellis files live in the workspace (~/.openclaw/workspace/), they're completely untouched by OpenClaw upgrades. Think of it like upgrading your OS — your documents stay the same.

In practice, we run Corellis on a production server with 28 agents and upgrade OpenClaw regularly (sometimes multiple times a week). Zero conflicts so far because the boundary is clean: OpenClaw = runtime engine, Corellis = what the agents know and how they behave.

The only edge case would be if OpenClaw introduces a breaking change to workspace file conventions, but that hasn't happened — they're good about backward compatibility.

[ralyodio]

i see maybe mention that in readme