BISKIN BRAIN · LOCAL-FIRST · CLAUDE-NATIVE

Your AI's long-term memory. Built in markdown, owned by you.

Every Claude session starts from zero. You re-explain the same context, paste the same links, rebuild the same mental model. Biskin Brain is an Obsidian vault + MCP bridge + nightly agent that gives Claude — Desktop, Code, mobile — a memory that survives every session. Local files. Claude-native. Permanent.

$79 $299
Three tiers
100% local
No cloud lock-in
Nightly
Sentience agent
Q2 2026
Gumroad launch

Claude forgets everything every time.

Every session starts cold

You've told Claude about your company 40 times. It still doesn't know. Each new conversation starts with the same context dump.

Memory files are read-only

`.claude/memory/` is static. Claude reads it, but never updates it, never cross-references, never notices contradictions.

Chat exports rot on disk

You've got Claude exports in Downloads going back 6 months. They're JSON. Nothing reads them. They might as well not exist.

PKM tools don't speak AI

Obsidian has 2,000 plugins. Notion has databases. Neither was designed for an agent to read, update, or reason over.

Cloud AI = your data is theirs

Every SaaS memory product wants your notes on their server, priced per-month, deprecated on their schedule. Not yours to keep.

Context windows don't scale

200k tokens is a lot. It's not a year of work. It's not your whole company. At some point, something has to live outside the prompt.

Four layers. Each one swappable.

L4 · AGENT

The Sentience Layer

Runs at 3am every night. Reads every note modified in the last 24 hours. Writes typed links between related notes. Updates Maps-of-Content. Flags contradictions with past claims. Writes a 300-word "state of the brain" each morning. ~$15/mo on Claude Opus 4.7 with prompt caching. Hard cost cap built in.

L3 · BRIDGE

The Claude bridge

obsidian-mcp-tools exposes your vault to Claude Desktop and Claude Code. Semantic search. Vault reads and writes. Templater access. Every Claude surface — desktop app, CLI, mobile — reads from the same brain.

L2 · INTEL

The intelligence layer

Smart Connections (local BGE-small embeddings, offline-capable). Obsidian Copilot for in-vault chat using your own Anthropic key. Templater + Dataview + Bases for the structure. Zero cloud dependencies unless you opt in.

L1 · STORAGE

The vault

Plain markdown files. Human-readable. Zero lock-in. Obsidian Sync ($8/mo) handles mobile. Nightly push to a private git repo handles backup. Triple redundancy: disk + OneDrive + git. If Obsidian dies tomorrow, you still have the files.

Three tiers. Pick the one that matches where you are.

SOLO
$79· one-time

For the DIY developer who wants the template and can wire the rest themselves.

  • Vault template (PARA + Zettelkasten structure, seeded MOCs)
  • All plugin configs (Smart Connections, Copilot, Templater, MCP)
  • Claude Desktop + Claude Code MCP setup JSON
  • Nightly Sentience agent Python script
  • Ingest pipeline scripts (Claude exports, Readwise, voice)
  • 3D graph visualization (Three.js, UnrealBloomPass)
  • 20-minute video setup guide
  • Lifetime updates
Get Solo →
BUNDLE
$299· one-time

For the serious user who wants the full Biskin Studio toolkit wired in.

  • Everything in Kit
  • All 7 Biskin Agents packs (founder, creator, gamedev, devs…)
  • Sample Biskin Creative case study setup
  • "Your AI's long-term memory" positioning deck (for your own pitches)
  • Quarterly 30-min check-in call for the first year
  • First access to v2 when it ships
  • Lifetime updates
Get Bundle →

The questions people actually ask first.

Do I need to be a developer? +

If you can install Obsidian and paste a file into a config folder, you can run Solo. The Sentience agent is Python — the Kit tier includes a 60-min walkthrough that handles it. If you've never touched a terminal, grab the Kit and I'll walk you through it on the onboarding call.

What does it cost to run? +

Obsidian Sync $8/mo (optional, but worth it for mobile). Claude API for the Sentience agent ~$15/mo for a typical vault (hard cap enforced in script). Smart Connections + Copilot = free tiers cover most people. Total ongoing: ~$23/mo. That's one coffee a week.

What if Obsidian or Anthropic goes down / deprecates? +

Your data is plain markdown on your disk. Obsidian dies, you open the files in any editor. Anthropic deprecates — the Sentience agent uses any OpenAI-compatible API. Vault structure is agnostic. Zero lock-in is the whole point.

Does this work on mobile? +

Obsidian mobile + Obsidian Sync — yes. Claude mobile can read the vault through the MCP bridge (if you run the server on a machine that's on). A fully mobile setup is on the v2 roadmap.

Can I try before I buy? +

The setup video is posted publicly on the devlog. The vault structure spec is in the README of the public repo. The Sentience agent is 300 lines of Python — glance at it. You're buying the packaged, tested, documented version + my time to fix it if it breaks for you.

When does it ship? +

Target Q2 2026. I'm dogfooding the vault right now — Solo tier ships first (end of May). Kit ships a week later. Bundle ships the week after. If you join the waitlist, you'll get launch-week pricing ($20 off each tier) and be first in line.

Refund policy? +

30 days, no questions. If it doesn't clear the problem stated at the top of this page, email me and I'll refund you same-day.

Join the waitlist. $20 off every tier at launch.

I'll email you when each tier goes live — Solo end of May, Kit a week later, Bundle the week after. Waitlist members get a 20-dollar launch-week discount on every tier, first access to the vault template a week before the public drop, and the setup video direct to your inbox.

~ No spam. Launch week notification + 20% discount code. Unsubscribe one-click.