Anyone else hacking on CARL and GSD?
I just spent the last ~36–48 hours trying to add it into an existing project setup. My main concern: I’m integrating it into a complex multi-repo project that’s already had hallucination + drift incidents—bad enough that I’ve lost weeks cleaning up corruption. So I’m being paranoid about how CARL gets introduced and gated.
One extra detail: I use a custom gsd:checkpoint command (I’m trying to get it committed to the GSD project; it’s currently marked low priority), and it’s been really effective at keeping drift under control.
Question for the group:
- How are you tying CARL into GSD safely?
- Are you routing specific GSD commands → CARL domains automatically?
- Has anyone built a clean pattern for command hooks / wrappers so CARL runs only when appropriate (and doesn’t amplify drift)?
- If you’ve got a working setup, what’s your minimum-safe integration approach?