A user from the LocalLLaMA community has expressed doubts about the actual usefulness of OpenClaw, a framework designed to simplify the use of large language models (LLMs). While acknowledging some advantages, the user suggests that the value lies mainly in the skills that are developed or integrated, rather than in the framework itself.

Memory and Automation

The user highlights how automatic memory management, one of the features offered by OpenClaw, can sometimes be counterproductive, "polluting" the context with irrelevant information. Alternatively, they prefer a "manual memory" approach, explicitly recalling the necessary information through specific prompts. Even the automation of tasks via cron, although useful, is considered less flexible than the ability to recall a specific skill on-demand, with updated data.

Skills vs. Runner

The user's conclusion is that the most important aspect in using LLMs lies in the development and integration of custom skills. In this context, tools like "opencode web" could represent a more effective alternative, shifting the focus from the execution infrastructure (the "runner") to the specific functionalities implemented.