claude-engram
Persistent memory for Claude Code. Auto-tracks mistakes, decisions, and context via hooks. Tiered storage, semantic scoring, multi-project workspaces. Local-only.
Claude Engram provides persistent memory and self-awareness for Claude Code, automatically tracking mistakes, decisions, and context across sessions via hooks. It includes loop detection, scope guards, tiered memory with archiving, semantic intent scoring, and optional local LLM for code analysis.
Features
Compatibility
Quick start
Use cases
Alternatives
Related searches
Comments
- Sasha RiveraMay 12, 2026
Persistent context means Claude Code improves with each session.
- Remy ChenMay 4, 2026
Auto-tracks mistakes, decisions, and context for Claude Code across sessions.
- KKendall GarciaMar 26, 2026
Automatic tracking is the right approach — no manual annotation required.
- EEllis LeeMar 14, 2026
Good for developers who want their AI to learn from past work on the project.