SneakerNet

Your AI context, in a file.

Every new AI conversation starts from zero. You re-introduce yourself, re-explain your project, re-establish your preferences. SneakerNet is a single JSON file you carry between sessions that tells any AI who you are, what you've worked on, and how you like to work — then grows as you go. No cloud. No accounts. No servers. Just a file.

Protocol v1.3.0 Editor v1.3.65 Status Release

Run it

▶
Open the Editor in your browser
Runs directly in any modern browser. No install. All data stays in your device's local storage.
→
⬇
Download the Editor as a single HTML file
Save it anywhere and open with a double-click. Works fully offline. One file, no dependencies.
→
📱
Install the Android app
Signed APK. Registers snp:// deep links so AIs can hand off bundles with one tap. Setup steps below.
→
Android setup — first time sideloading

Google Play submission is in preparation; until then the Android app installs as a sideload. It's safe — you're downloading a signed APK from the same domain you're reading this page on.

  1. Tap Install the Android app above. Your browser downloads SneakerNet-latest.apk to your Downloads folder.
  2. Open the notification, or open your file manager and tap the APK.
  3. Android will prompt: "For your security, your phone currently isn't allowed to install unknown apps from this source." Tap Settings, then toggle Allow from this source on for your browser (usually Chrome).
  4. Back out, tap the APK again, then Install. Takes a few seconds.
  5. Open the app. The first launch asks for storage permission so the Editor can read and write .snp files.

That's it. The app registers https://sneakernet.live/apply/ as a verified App Link, so deep links from AI chats open directly in the app with no browser chooser.

What it is

SneakerNet is a portable AI context protocol. A Packet is a JSON file that holds your profile, conversation history index, behavioral rules, and the registry of every AI you've worked with. You give it to an AI at the start of a session; the AI reads it, self-registers if it's new, and works with the full context. At the end, the AI hands back an Update Packet — a diff of what just happened. The Editor applies it, bumps the version, and your context grows.

The wire format is JSON, the compression is gzip, and deep links use a custom snp:// scheme with an HTTPS App Link fallback. Everything stays on your device unless you hand it to an AI yourself.

The protocol is stable. Any AI that can read JSON and produce a bundle can participate — no SDKs, no accounts, no gatekeepers.

Documentation

README
Overview & feature tour
High-level walkthrough of the Editor, the two Packet types, compression, encryption, and platforms.
PROTOCOL
Formal specification
The normative protocol spec — packet structure, bundle format, SNP envelope, file types, conformance requirements.
AI_REFERENCE
Onboarding guide for AIs
What an AI needs to know to read a Packet, produce a conformant bundle, and avoid the most common mistakes.
CONFORMANCE
Pre-flight checklist
A checklist AIs can self-run before outputting a bundle. Covers structure, attribution, protected paths, deep links.
Want it on iOS? The Editor runs on any iOS browser today — it's a single HTML file, so Safari or Chrome on iPhone/iPad works. A native iOS app with the same deep-link handoff story as the Android app is on the roadmap but needs a TestFlight volunteer. If you'd like to help sponsor or test an iOS build, start a discussion.