memory-ts

Memory-ts preserves conversation context, debugging insights, and architectural decisions across Claude Code and Gemini CLI sessions so your AI assistant remembers previous work and picks up where you left off. Using semantic embeddings and a multi-signal activation algorithm, the system automatically surfaces relevant memories based on meaning rather than keywords. Installation takes one command, the memory server runs alongside your coding sessions with zero friction, and curation happens automatically at session end. Developers benefit from continuous context without re-explaining decisions or rebuilding rapport with each new session. The system handles concurrent sessions from different CLIs seamlessly and supports both global cross-project memories and project-specific technical details.

Install globally with `bun install -g @rlabs-inc/memory`, then run `memory install` for Claude Code or `memory install --gemini` for Gemini CLI. Start the memory server with `memory serve` and verify setup with `memory doctor`. Use your AI assistant normally—memories extract and surface automatically.