Grok build is pretty freaking awesome. Small issue with memory for more complex projects, so I integrated it with my Claude Code memory systems.
Built a full automatic memory system for Grok Build using SLM + claude-mem + Graphiti. Automatic project scoping, read enrichment on every prompt (SLM recall + live Graphiti facts via socket), and write capture on Stop with clean user+assistant pairs, context-preserving outcome extraction, direct claude-mem writes, and chunked Graphiti episodes. Fail-silent with strong recency bias.
@seriousposterr I had this and realized the cron was spamming my conversation. It was taking Most of the context. I removed all cron from OpenClaw and created a separate tool to manage any cron like needs for openclaw. No issues after that.
I just created a service that would monitor all of my tools and make sure they were running. I only provide Moltbot one interface. That way we’re not using the Moltbot cron. Which also fills up the lines sent to the AI query! I was getting nonsense back from my chats and realized my cron usage was filling up the context!
It also has an API to run all tools so the https://t.co/EGNv8lvCOH is super simple. One ring to rule them all!
@openclaw with @xai is a game changer. Larger context window with 2 million tokens and a stricter adherence to MD file. So much better than the alternatives. You know who you are.