Productivity
obsidian-mind
A self-organizing Obsidian vault that gives AI coding agents persistent memory across sessions.
★ 5.0k starsMIT
View guide →Persistent memory for Claude Code and other CLI agents, stored as plain Markdown in your Obsidian vault
obsidian-second-brain is an open-source productivity project by eugeniughelbur (eugeniughelbur/obsidian-second-brain on GitHub). It has 4,697 GitHub stars and 580 forks, is released under the MIT license, and was last updated on GitHub on 7 October 2026.
Use an existing Obsidian vault or create a fresh one to serve as the memory store.
Clone the repo and follow its install instructions for your agent (Claude Code first-class; other CLIs supported).
Run the init command to set up the vault structure and pick a role preset (developer, founder/operator, writer, or researcher).
Capture a conversation with the save command and recall it in a later session with the find command.
Commands for saving, ingesting, finding, synthesizing, reconciling, daily notes, tasks, people, calendar, and more.
New sources update existing pages instead of appending duplicates; contradictions are reconciled automatically.
URLs, PDFs, audio (transcribed locally with Whisper), and screenshots are distributed across entity pages, task boards, and daily notes.
Four scheduled agents — morning brief, nightly consolidation, weekly review, and vault-health check — maintain the vault unattended.
Commands pull from X, the web, and YouTube with cited dossiers; vault-first deep research fills only the gaps in what you already know.
Memory lives as Markdown in your Obsidian vault and can be exported as a snapshot any AI tool can read.
Clone the project and enter its directory.
git clone https://github.com/eugeniughelbur/obsidian-second-brain.git
cd obsidian-second-brainFollow the README's install section for Claude Code or your CLI agent, so the /obsidian-* commands become available.
# see README install section for the exact command for your agentInside your agent, initialize the vault, then capture and recall memory.
/obsidian-init
/obsidian-save
/obsidian-findEugeniu Ghelbur, an AI automation engineer; the project builds on Andrej Karpathy's LLM Wiki pattern.
Yes, MIT-licensed. You use it with your existing agent subscriptions; the README emphasizes key-less web research, and transcription uses a local Whisper install.
No separate key is required by the skill itself — it runs inside agents you already have (such as Claude Code), and several research commands are described as key-less by default.
Those typically add chat or search inside Obsidian. This skill makes the vault the agent's long-term memory across CLI tools, with self-rewriting pages and scheduled maintenance agents.
Clone the repo, install it for your agent per the README, run `/obsidian-init`, then try `/obsidian-save` and `/obsidian-find`.
A self-organizing Obsidian vault that gives AI coding agents persistent memory across sessions.
An Obsidian dashboard for your Claude Code sessions.
Open Brain (OB1): one database, one AI gateway, one capture channel — shared memory for every AI you use
Anthropic's plugin suite for legal workflows — practice-area plugins, named agents and MCP connectors.
A Claude Code skill that writes content in your voice.
5 no-cost Claude AI skills for Product Managers & Business Analysts — PRDs.
The most-starred other Productivity projects in the LibreAI catalogue — the realistic shortlist if obsidian-second-brain is not the right fit.
Real-time global intelligence dashboard: AI news aggregation, geopolitics and infrastructure tracking
Memos — a personal timeline for quick notes: capture short memos in seconds, find them later by search, tag or date; open source and self-hosted
Joplin — the privacy-focused note-taking app: Markdown notes and to-dos on every platform, offline-first, with end-to-end encrypted sync you control
Open-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, mult…
Logseq — a privacy-first platform for knowledge management: an outliner over your local Markdown files, with a graph, journals and spaced repetition built in
Free and open — grab the source, star the repo, and start creating.