How do OpenClaw and Remember coordinate their memory retrieval and writing processes?

[liang9886703]

I installed remember using the standard procedure outlined in the documentation, like this:

openclaw plugins install @remember-md/remember
openclaw gateway restart

Afterwards, I communicated with Openclaw, telling it:

/remember:init

Created in the default path

Serves the current project

/remember:process

At this point, I'm unsure if this conversation will automatically write to and retrieve contents under ~/remember (my workspace is obviously not there).

[codezz]

Hello, so the extraction process is not fully automated in terms of running after each session unit, so you need to run the /remember:process command. This can of course be configured as a cron inside OpenClaw.
Let me know if it works.

[liang9886703]

I saw the /remember skill, and I added the following to AGENTS:

• If you think it's appropriate to use the remember skill, simply record the information and notify that it has been recorded. Execute it directly without prompting.

This should replace running the /remember:process command.

Now, I'm unsure if the ~/remember workspace will be retrieved by Openclaw as a memory.

[liang9886703]

I found that relying solely on prompt words wasn't stable enough, so I switched to using cron as a backup: a scheduled task called remember:process has been created to run automatically once a day at 5:00 AM. It seems to be working well so far.

[codezz]

Great. Thanks for sharing.

[kneelgiee]

Love this, but anyone here know at some point any openclaw wrapper that has persistent memory? setting up stuff to openclaw has been very tiring recently:)

[codezz]

One thing I found recently is that you can add extra paths where openclaw can search other that it's default memory. You can add the second brain path there