lucasrosati/claude-code-memory-setup

Up to 71.5x fewer tokens per session on Claude Code with Obsidian + Graphify. Persistent memory, codebase knowledge graphs, and chat import pipeline. 🇧🇷 PT-BR included.

What it solves

This project provides a comprehensive setup to give Claude Code persistent long-term memory and codebase awareness. It solves the problem of "amnesia" between sessions—where users must re-explain project details—and the high token cost associated with the AI re-reading entire codebases to understand structure.

How it works

The system uses a three-layer approach to manage context:

  1. Project Memory (Obsidian): A centralized Obsidian vault acting as a "second brain" using the Zettelkasten method. It stores decisions, progress, and knowledge. Claude Code interacts with this vault via a CLAUDE.md instruction file and custom /resume and /save commands.
  2. Code Map (Graphify): It integrates Graphify to transform codebases into queryable knowledge graphs. This allows Claude to query structural information (via AST) rather than reading raw files, significantly reducing token usage.
  3. Chat Import Pipeline: A Python-based pipeline that exports Claude conversations (from both the CLI and Web app) and imports them into the Obsidian vault as tagged, interlinked notes.

Additionally, a safety-net auto-save hook is provided to mechanically extract session facts from transcripts and log them to the vault upon session end.

Who it’s for

Developers using Claude Code who want to maintain a persistent knowledge base across multiple sessions and optimize token consumption when working with large codebases.

Highlights

  • Persistent Memory: Uses Obsidian for declarative memory (decisions, architecture).
  • Token Efficiency: Leverages Graphify for structural codebase mapping to avoid redundant file reads.
  • Automated Sync: Includes a pipeline to convert chat histories into searchable vault notes.
  • Zettelkasten Integration: Implements a structured note-taking system for AI-assisted development.
  • Session Continuity: Custom commands and hooks ensure that progress and context are preserved automatically.

Related

  • Project
    supermemoryai/claude-supermemory

    claude-supermemory is a Claude Code plugin that gives the LLM coding agent persistent memory across sessions. It automatically captures conversation turns, recalls relevant past context at prompt time, and shares that memory across tools like Codex and OpenCode via a unified container tag.

  • Project
    breferrari/obsidian-mind

    A template vault for Obsidian that adds persistent, searchable memory to AI coding agents (Claude Code, Codex CLI, Gemini CLI) via hooks, slash commands, and optional local semantic search (QMD).

  • Project
    letta-ai/claude-subconscious

    A background agent plugin for Claude Code that provides persistent memory and proactive guidance by observing sessions and exploring the codebase via the Letta Code SDK.

  • Dispatch
    Maximizing Claude Code Session Value and Token Efficiency

    Optimizing Claude Code sessions requires strategic use of prompt caching, context management, and subagents to reduce token costs and prevent cache misses.

  • Project
    EliaAlberti/cpr-compress-preserve-resume

    A Claude Code plugin that provides persistent memory across sessions by compressing conversations into curated logs and automatically restoring context.