Obsidian Mind
A ready-made Obsidian vault that gives your command-line coding agent durable memory. When a session opens, hooks automatically inject your North Star, active projects and recent notes into the context of Claude Code, Codex CLI or Gemini CLI. You talk about your day, the agent writes the notes, refreshes the indexes and maintains the links by itself. Eighteen built-in commands cover standup, weekly review or incident capture. The bridge between agent and vault rests on five lifecycle hooks: session start (context gets loaded), prompt submit (the message gets routed), after each tool call (the note just written gets checked), before compaction (a backup is taken) and stop (a checklist). Reading happens in tiers: light context at startup, semantic search on demand, and a full note read only when it is actually needed, so the whole vault never lands in the window.
Strengths
- Memory injected automatically at startup: the agent re-reads your North Star and projects without you pasting context
- Works with three CLI agents (Claude Code, Codex CLI, Gemini CLI), not locked to a single vendor
- Semantic search via QMD and a wrap-up command that verifies notes and rebuilds the indexes
Limitations
- Heavily opinionated structure: you take its tree and commands wholesale, hard to keep just one piece
- Without prior Obsidian practice, the setup effort outweighs the benefit versus a plain notes folder
Best for
- Devs already living in Obsidian who want their CLI agent to pick up a project thread without re-explaining
- Anyone wanting to test the Karpathy pattern by cloning a full example rather than starting from a blank page