sno-ai/sno-station v1.5.0 is a release that changes the ordering of memory search (Recall) results. A small ranking model running on the user's own device moves the memory that corresponds to the answer to the top of the results. The model reads English and Chinese to assign ranks. On the first call it downloads the model in the background, and existing behavior is kept until the download finishes.
sno-ai/sno-station is a project in which Codex, Claude Code, Cursor, Hermes Agent, and OpenClaw share a single encrypted memory and hand off or review work for one another. Instead of each agent relying on the memory it holds separately, it is configured to collaborate on the basis of that shared memory. According to the project description, it is open source and runs on the user's computer. The change in this version does not concern that shared memory itself, but the stage that orders memory search results.
The model download follows the proxy settings, resumes from where it stopped, and automatically re-downloads corrupted model files. However, the languages the ranking model reads are given as English and Chinese, so the given information says nothing about how Korean memories are ranked. No figures are given either for how much the reranking improves search results, so the effect of this change is hard to judge beyond the feature description.