Second Brain logo

Second Brain

Private tool AI / Knowledge

Second Brain is a local Markdown vault that gives my AI tools a shared reference between sessions. Notes stay on my machines and synchronize peer to peer.

My role. I designed the Markdown structure, reviewed context packs and peer-to-peer sync so my AI tools can pick up work across sessions.

See a memory exampleExplore Second Brain VR

From one session to the next

A fictional project shows how context moves between sessions without exposing private notes.

  1. Record a finding

    Session noteThe demo needs a clearer first screen.

    An agent records the finding in a new inbox note with its source.

  2. Review and curate

    Maintained contextKeep the demo introduction concise.

    During curation, reviewed information can be incorporated into the project notes and pack.

  3. Resume the work

    Next sessionRead the relevant project pack

    A later session reads the maintained context before continuing the work.

Design decision I separated new session notes from maintained references so context can be reviewed before it guides later work.

The problem

New AI sessions need context. A maintained reference helps me avoid repeating the same project history.

How it works

  1. Read An agent reads the relevant project reference.
  2. Record Session findings are kept separate from canonical notes.
  3. Curate Reviewed information becomes maintained notes and packs.
  4. Sync Syncthing keeps the machines aligned.

What it includes

  • Plain Markdown files
  • Peer-to-peer synchronization
  • Curated project packs
  • Local review dashboard and backups

Second Brain VR

WebXR prototype · headset validation pending

A read-only spatial view of a local knowledge graph, built for exploration in the Meta Quest 3S browser.

  1. Read a graph

    Open a read-only representation of connected notes.

  2. Explore in space

    Navigate spatial nodes and panels in a WebXR environment.

  3. Test the experience

    The implementation is available for development testing. Physical headset validation is pending.

Project status

Active as my private project memory. The vault and its contents are not published in this portfolio.

Technical details

Markdown and Obsidian, Syncthing with Tailscale, and local Python automation. Cloud agents can read the context required for a session; this is not a claim of zero cloud processing.

  • Obsidian Markdown
  • Syncthing (Local P2P)
  • Python Automation
  • Tailscale
  • launchd
Back to top