diego-ninja/memento

Persistent memory for LLMs with lossless transcript management. Two-layer architecture: knowledge graph (decisions, learnings, preferences) + full session history with hierarchical summary DAG. Real-time capture, regex search across all past conversations, and compaction-aware context recovery. MCP server for Claude Code.

2 files for Claude Code: settings, memento CLAUDE.md — 299 tokens loaded in every session.

1Stars on the repository
2Files it configures its agents with
299Tokens loaded in every session
1Agent configured

Instructions

Settings

These files are diego-ninja/memento's own configuration — they tell Claude Code how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.