Obsidian is becoming the long-term memory for all my agents

Start Here
,
1

I use Obsidian as a centralized long-term memory for my agent apps. Mostly my OpenClaw agents, but also Codex, Claude, and even Hermes. Having them work with the same markdown notes makes projects much easier for me to track across apps.

Their Obsidian skill includes instructions to keep notes visually readable, with visual cues so I can open a file and quickly understand what’s going on. I’m going to be reading these myself :grinning_face_with_smiling_eyes:

Still convinced human+AI readable markdown is the supreme intelligence.

How do you organize this when several agents use it? Multiple vaults, or one vault to rule them all? And inside the vault, do you split things by project or by agent?

2

How does this compare to the build in memory systems of openclaw?

3

I’ve used Obsidian for about 3 years before OpenClaw ever came into the picture. One vault to rule them all - folders to isolate different spheres of knowledge - links between disparate areas. I maintain the vault alongside my agent.

1 Like