Mindnook provides isolated Firecracker microVM hardware sandboxes combined with an interactive Obsidian knowledge graph, giving your autonomous AI agents a secure environment to research, draft, code, and organize your thoughts without token markups.

How does Mindnook connect with Obsidian?

Every insight, research note, and task generated by your AI agents is persisted as native 100% Markdown files with [[wikilinks]]. You can export or synchronize the entire vault directly into Obsidian on your desktop.

How much does a Dedicated MicroVM Box cost?

Mindnook offers two dedicated compute tiers: Interactive Sandbox at $19/month (with smart sleep) and Always-On Autonomous Box at $39/month (24/7 compute for background cron tasks). Both tiers include zero token markup.

Do I have to pay per-token markups on AI models?

No. Mindnook charges flat pricing only for isolated hardware compute ($19/mo or $39/mo). You can plug in your own API keys for Claude, OpenAI, Gemini, or connect open-weights models running on your local machine.

Dedicated MicroVM Cloud · Obsidian Second Brain · Native Desktop Ready

a quiet room for a
crowded mind.

Your thoughts mapped into an interactive Obsidian knowledge graph, while your autonomous AI crew does the heavy lifting 24/7 inside a private Firecracker microVM.

Scout Deep Research

Scribe Vault Architect

Radar Market Intel

Errand Ops & Systems

Tock Routines & Cron

Stack Deep Focus

Zero Overhead Autonomy

You sit around. Your bots do the rest.

Pour a warm cup of coffee while your agent fleet crawls literature, maps your Obsidian graphs, watches onchain liquidity, and compiles markdown digests in parallel.

😌

☕

You: Resting & sipping coffee

Obsidian Second Brain 6 nodes · 8 links

Every research query, paper summary, and synthesis is automatically mapped into linked Markdown nodes with bidirectional [[wikilinks]].

Scout (Deep Research)

Running in MicroVM Sandbox

2 vCPU · KVM

Analyze the new continuous-depth Liquid neural networks and link it to our edge AI vault.

Scout (Deep Research)

Paper analyzed. Liquid foundation models eliminate transformer quadratic attention for linear-time edge inference.

curl arxiv.org/abs/2402.05141 200 OK

Obsidian Note Ready:
[[Liquid Neural Networks]] ↔ [[Edge LLM Inference]]

Autonomous Office Maze

Always on patrol. Each in their own lane.

Your agent crew navigates dedicated corridors across your digital headquarters 24/7, keeping operations, research, and vaults constantly in sync.

Active Autonomous Corridors · Click Any Agent

The Architecture

Dedicated hardware. Not another API wrapper.

Every Mindnook box is a dedicated Linux environment running on Fly.io with native desktop syncing.

Private MicroVM Box

Hardware-isolated Firecracker microVM with dedicated 2 CPU cores, 2048 MB RAM, local SSD, and a headless browser for autonomous web navigation.

Hardware Isolated

Obsidian Second Brain

No lost knowledge in ephemeral chats. Every finding is indexed as native Markdown with [[wikilinks]], ready to export and open inside Obsidian at any moment.

100% Markdown Storage

Hermes Desktop Sync

Connect directly to your macOS or Windows native Hermes Desktop app via secure WebSockets. Control your cloud box straight from your desktop dock.

Native Desktop Ready

Two Plans. Dedicated Firecracker Compute.

Whether you need on-demand interactive deep research or continuous 24/7 background agent patrols, every box runs in an isolated Linux microVM.

Live Soon
Interactive Sandbox

$19 / month

Ideal for ad-hoc deep research, Obsidian note-taking, and interactive PWA & Hermex chat sessions with smart sleep.

2 vCPU · 2048 MB RAM Firecracker MicroVM
3 GB Persistent NVMe Vault (/root/data)
Smart Sleep (Scale-to-Zero) after 10m idle
Instant Cold Start (< 1.2s) on any prompt
Obsidian Thought Graph automatic sync
Zero Token Markups (Bring Your Own Key)
Hermex iOS Deep-Link & PWA 1-click install

Why Mindnook Dedicated Boxes?

Compare running local AI agent daemons on your laptop vs. an isolated Mindnook Firecracker MicroVM.

Architecture Spec	Mindnook Cloud Box	Local Laptop Gateway	Standard AI SaaS
Compute Isolation	Dedicated KVM MicroVM	Shared Host OS	Shared Multi-Tenant Pod
24/7 Autonomy	Always On (Zero Battery Drain)	Stops when laptop sleeps	Limited by credit burns
Knowledge Storage	100% Native Markdown Vault	Local file system	Proprietary Cloud DB
Obsidian Second Brain	Automatic [[wikilinks]] Sync	Manual export	Not supported
Model API Pricing	Zero Token Markup (BYOK)	Zero markup (BYOK)	3x - 10x Token Markup
Desktop Integration	Live WebSocket Protocol	Tailscale / Cloudflare Tunnel	Browser only

Frequently Answered Questions

Everything you need to know about Mindnook MicroVMs, Obsidian integration, and billing.

Mindnook provides dedicated Linux compute microVMs running on AWS Firecracker KVM technology, paired with an interactive Obsidian knowledge graph. Your autonomous AI agent personas (Scout, Scribe, Radar, Ops, Tock) work in the background to conduct research, organize thoughts, and build persistent second-brain vaults.

Every note, paper breakdown, and synthesis generated by your agents is saved as 100% native Markdown with bidirectional [[wikilinks]]. You can export the files directly or sync them straight into your local Obsidian application via git or direct volume download.

No. Mindnook operates on a pure compute model ($15/month). You provide your own API keys for Claude (Anthropic), OpenAI, Gemini, or connect to local open-weight LLMs via Ollama with zero markups or token taxes.

Each dedicated box runs 2x vCPU, 2048 MB RAM, dedicated persistent NVMe storage, isolated networking, pre-installed Python/Node toolchains, and a real-time WebSocket endpoint compatible with the native Hermes Desktop application.

We support all major Credit Cards, Apple Pay, Google Pay, and PayPal via Paddle (Merchant of Record) and Lemon Squeezy, as well as zero-KYC crypto payments (USDC and SOL).

Welcome to Mindnook

a quiet room for a crowded mind.