Jevmem

•
155
•

Automatically captures project decisions, constraints, bugs, and todos from AI coding assistants to improve context recall.

Jevmem acts as an automatic project memory for AI coding assistants like Claude Code, Cursor, and Codex. It intelligently monitors your coding chat sessions, automatically saving key decisions, constraints, identified bugs, and pending tasks into a structured `JEVMEM.md` file. When information changes, old entries are marked as superseded rather than deleted, maintaining a complete project history. Crucially, before starting a new session, Jevmem injects the most relevant historical context back into the AI's memory, ensuring that your AI assistant always has the necessary background. It also includes robust security features like secret scrubbing and a memory-poisoning check to validate externally added memory lines.

Key Features

01Automatically captures decisions, constraints, bugs, and todos from AI coding assistant chats
02Performs a memory-poisoning check on external or hand-edited memory lines before injecting into AI context
03Automatically recalls relevant project context to AI assistants for each new session
0492 GitHub stars
05Automatically scrubs common secrets (e.g., passwords, API keys) from chat turns before processing
06Supersedes outdated memory lines instead of deleting them, preserving historical context

Use Cases

01Preventing redundant explanations and memory loss when interacting with AI coding tools
02Ensuring AI agents have full context of past decisions and constraints across coding sessions
03Maintaining a persistent, evolving project memory for AI coding assistants