Memory Systems & Building Long-Term Context
Questions
QMD as better memory solution? Yay or nay?
Graph + Vector DB as memory, any standardised approach?
How much, roughly do you spend on opus if you are comfortable sharing. If not, tips for getting your cost down.
How to transfer memory from chat/Claude/gem to Open Claw?
Can you share any approaches to weighting context?
Could you please post your sleep memory system on GitHub?
How do you store the "why" of the memory?
How to you do remove / prune irrelevant memory? And have the systems learn?
What is the difference between cron job and heart beat
How can you avoid “memory bloat”
Isn’t rag a sufficient solution for most tasks with skills and projects
Why is context limited? How to expand?
How is playwright different from when openclaw uses the chrome browser?
Tips on stopping our models overthinking?
Do you find “skills” limiting?
Do you find “skills”
Ever experimented with a local uncensored model?
Is langraph dead?
When should you use memory and core files vs skills as memory? (Can explain more if you ask this)
Does Obsidian work as a Knowledge Base if we have indexed the MD files?
What are some good open source models to use? Can you set them up on digital ocean?