Show HN: I turned Claude Code into a personal assistant

9 pointsposted 5 hours ago
by daxaur
(github.com)

5 Comments

auvira_systems

5 hours ago

I love seeing more 'Local-First' tools hitting the front page. We’ve been moving our own project (a financial IDE) toward a purely local-first architecture, and the biggest challenge is always balancing zero-cloud with useful features. Using Claude Code as the base is a clever move because it already has the context of the filesystem. Have you looked into adding a local vector DB (like LanceDB or similar) to give 'pawmode' long-term memory without needing an external API?

daxaur

4 hours ago

Thank you!

Memory right now is flat markdown files (MEMORY.md , SOUL.md) that Claude reads at session start. Honestly it works better than you'd expect. Claude's context window is big enough that a few hundred lines of memory covers most personal assistant use cases. I'd rather keep it as plain text you can read and edit yourself than add a vector DB dependency for marginal recall gains.

The filesystem context thing is exactly right btw

stuaxo

4 hours ago

Going to object to anything using Claude as being local.

Still: this totally could be hooked up to a local LLM.

mittermayr

4 hours ago

OpenClaw vs. OpenPaw? Seems like a very similar thing, or am I misunderstanding the concept here? Is the lock-in to Claude Code the feature here?

Can't be long until we see a BaseClaw repo, where everyone can clone and name their own local claw-like loop.

daxaur

4 hours ago

It is indeed very similar, but

OpenClaw doesn't support Claude Code's native OAuth, plus it runs its own daemon. OpenPaw goes the other direction: it writes config files that Claude Code already understands (CLAUDE.md, permissions, hooks) and exits so claude code can function as an actual personal assistant with all the skills needed. No daemon, no extra process.