Browse / Second brain / obsidianbrain
obsidianbrain
v-khanna
My take on an Obsidian + Claude Code second brain.
Open v-khanna/obsidianbrain → Checked against the four rules.
ai-agentsclaude-codeobsidianpersonal-knowledge-managementpkmsecond-brain
Built for one agent
ships CLAUDE.md and nothing for other agents.
In its own words
Skills do the filing.
How I built an Obsidian setup that mostly takes care of itself. I got tired of pretending I'd keep up with my own system, so I let a bunch of Claude Code skills do the filing, journaling, and audits. I write into an inbox, the skills move things to the right place, and I read the vault in Obsidian like a normal person.
What's different from a generic Obsidian setup
Most "second brain" templates are scaffolding plus discipline. You set up PARA or Zettelkasten, then spend six months losing to your own system. I stopped trying. Here's what I did instead.
The skills do the actual work. Not Templater. Not Dataview-as-database. There's a /brain-triage that reads my inbox and files things to the right workspace. A /brain-record for journaling as I go. A /brain-lint that flags orphans and dead links. They're regular Claude Code skills, invoked with a slash command. The vault is shaped around what they need.
Top-level folders are domains, not buckets. PARA never stuck — I'd put the same idea under "Area" and "Resource" and "Project" and lose track of which was the canonical one. So the top-level is just whatever your life looks like: Work/, Career/, Projects/, Personal/, whatever. Each one is its own workspace with its own rules.
The repository's own description, then 19% of its README. Read the rest on GitHub ↗
Read from the repository's git tree on 2026-08-17.