Codelux Tool Gives AI Coding Agents Persistent Memory Across Sessions
A developer has built Codelux, a local daemon tool designed to solve the problem of AI coding agents losing all project context at the end of each session. The tool indexes a codebase locally in seconds, storing files, symbols, call flows, and developer decisions in a local database that persists between sessions. When a new session begins, the agent queries this stored memory instead of re-reading entire files, which the creator claims can reduce token usage by up to 90% on individual file reads. A Team plan allows multiple developers or agents on the same project to share decisions and notes in near real-time with end-to-end encryption. Codelux works via the MCP protocol on localhost and is compatible with tools like Claude Code, Cursor, Codex, and VS Code, with a free tier available for one project per machine.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in