The blog
Field-tested tips for every agent package — what to set up first, why it matters, and what good looks like.
OpenClaw's REM (Reflective Experience Memory) system lets your agents consolidate knowledge between conversations — promoting short-term observations into lasting expertise.
We just shipped a unified Memory & Context section in your dashboard. Here's what each setting does, which ones to enable, and the recommended configuration for most users.
Default vector search misses exact keywords. QMD adds BM25 full-text search, vector similarity, and LLM re-ranking — all running locally on your instance with zero API keys.
Standard context windows lose old messages as conversations grow. Lossless-Claw preserves every message using DAG-based summarization, so your agent never loses track of what was said.
Want these in your inbox?
Every release in plain English, once or twice a month — plus posts like this one.