Jevmem – automatic project memory for Claude Code, built on Jev (github.com)

A new tool called Jevmem has been announced, designed to automate project memory for Claude Code and compatible with Cursor and Codex. This innovative solution captures crucial information—such as decisions, constraints, bugs, and to-dos—from user interactions in real time, storing them in a file called JEVMEM.md. When users modify a decision, the previous line is marked as "superseded" rather than deleted, facilitating better tracking of project changes. Jevmem's integration allows relevant historical context to be injected into future sessions seamlessly. Significantly, Jevmem enhances productivity within the AI/ML community by streamlining how developers manage project memory. With high accuracy rates in decision-making—98.5% save/skip and a rapid response time of 0.30 seconds—Jevmem illustrates a compelling alternative to traditional large language models (LLMs) in terms of cost and speed. This tool promises to reduce cognitive overhead for developers and improve the efficiency of iterative coding tasks, while its design minimizes data retention and ensures that sensitive information is scrubbed before transmission. Overall, Jevmem represents a substantial leap forward in managing system memory in collaborative coding environments.