Descriptions:
Bart Slodyczka shares seven ready-to-use prompts designed to make OpenClaw agents significantly more capable and cost-efficient. The centerpiece is the QMD memory upgrade, which replaces OpenClaw’s default keyword-based memory system with a local RAG pipeline featuring vector search, semantic matching, and a reranking layer — all installable on a Raspberry Pi or Mac Mini. Slodyczka tested this on a Raspberry Pi with 16GB of RAM using Sonnet 4.6, which automatically resolved an installation error mid-process and updated the configuration to set the backend to QMD mode.
Other prompts include a daily morning brief powered by Tavily Search (which offers 1,000 free credits per month, replacing Brave API after its free tier was discontinued), an accountability partner agent that breaks user-defined goals into daily check-ins and weekly reviews stored in a local database, and a dashboard generator for visualizing progress without spending extra tokens on queries. Additional prompts give the agent the ability to self-update its instructions based on conversational feedback, safely stage software updates before applying them, and audit its own context files to reduce per-message token overhead.
The video targets users already running an OpenClaw setup and recommends Sonnet 4.6 or Opus 4.6 for the more complex installation prompts. All seven prompts are available as a free download on Slodyczka’s website.
📺 Source: Bart Slodyczka · Published March 07, 2026
🏷️ Format: Hands On Build
