breferrari/obsidian-mind

A self-organizing Obsidian vault that gives AI coding agents persistent memory. Claude Code, Codex CLI, Gemini CLI.

What it solves

AI coding agents typically lack persistent memory, starting every session from zero and forcing users to re-explain goals, team patterns, and past decisions. Obsidian Mind provides these agents with a "brain" in the form of an Obsidian vault, allowing knowledge to compound across sessions and machines.

How it works

It transforms an Obsidian vault into a persistent memory layer for agents like Claude Code, Codex CLI, and Gemini CLI. The system uses a combination of procedural hooks and agent-driven content creation:

  • Lifecycle Hooks: Node.js scripts handle automatic tasks such as injecting "North Star" goals and active project context at session start, classifying user prompts to route information (e.g., identifying a "win" or a "decision"), and validating markdown writes to ensure proper linking and formatting.
  • Tiered Context Loading: To manage token costs, it avoids dumping the whole vault into the prompt. It uses lightweight excerpts and file listings for initial context, and on-demand retrieval for specific details.
  • Semantic Search (via QMD): Optionally integrates QMD to provide local, model-based semantic search (using embedding, query expansion, and reranker models) so agents can find relevant notes even without keyword matches.
  • MCP Server: An Model Context Protocol (MCP) server allows agents working in any repository to search, read, and write to the central memory vault.

Who it’s for

Developers using AI coding agents (specifically Claude Code, Codex CLI, or Gemini CLI) who want their AI to remember project decisions, personal preferences, and professional wins across different coding sessions and repositories.

Highlights

  • Agent-Agnostic Core: Works across multiple CLI agents using shared Markdown and TypeScript hooks.
  • Automated Memory Management: Includes commands like /om-standup for morning kickoffs and /om-dump for rapid brain-dumping of meeting notes.
  • Local Intelligence: QMD integration enables offline semantic search without API costs.
  • Cross-Repo Access: The MCP server enables a single "brain" vault to serve as the knowledge base for all projects on a machine.

Related

  • Project
    jaredrhod/ai-memory-vault

    A system that turns an Obsidian vault into a persistent, external memory for AI agents, allowing them to remember information across sessions without context limits.

  • Project
    eugeniughelbur/obsidian-second-brain

    An AI-powered persistent memory system for Obsidian that allows AI assistants to read, write, and automatically maintain a self-evolving knowledge base.

  • Project
    rohitg00/agentmemory

    A persistent memory server for AI coding agents that automatically captures and retrieves project context across sessions using hybrid search.

  • Project
    coddingtonbear/obsidian-local-rest-api

    An Obsidian plugin that runs a local HTTPS REST API and a Model‑Context‑Protocol server, letting scripts, browser extensions, and AI agents perform full CRUD, targeted markdown patches, searches, command execution, and more on your vault. Authentication is via a bearer token and a self‑signed certificate; the plugin also provides ready‑made MCP configs for Claude and Cursor.

  • Project
    itechmeat/open-second-brain

    An Obsidian-native memory layer for AI agents that stores knowledge as plain Markdown files, allowing users to own and edit their agent's memory.