innernet docs

One memory for every AI tool. innernet.live holds it, netti keeps it current, and every AI you connect reads and writes it live over MCP. No copy-paste between sessions, no re-explaining yourself to every new chat, nothing to maintain by hand.

New here and not technical? The starter guide walks the whole thing step by step, with pictures of every screen. These docs are the other half: how it works, and everything you can build on it.

three ways in#

npx innernet signs you in and wires every AI tool on your machine (Claude Code, Cursor, Codex, Windsurf, Gemini CLI, Claude Desktop, VS Code, Cline, Zed, and more). Run it inside a repo and that folder gets a memory of its own. Get started
One URL. Paste https://innernet.live/api/mcp into any MCP client and sign in once in the browser. ChatGPT and Claude.ai connect the same way. Connect an AI
REST API. Build apps and agents on the same memory: the same 40 handlers, over HTTPS. API guide

what you can do with it#

Ask it anything. Every connected AI calls innernet_context with your question and gets back the few sections that answer it, inside a hard budget, instead of the whole cabinet. How retrieval works
Never re-explain yourself. innernet keeps a separate memory of you: your tools, your preferences, how you work. Each fact carries a disclosure class, and the private ones stay inside innernet unless you allow an app. Personal memory
Keep code and context together. Every commit lands as a capture; netti folds it into the memory; a teammate who clones the repo inherits it. Repo memory
Try an idea without losing the old one. Branch the whole memory, diff it against trunk, merge it back three-way. Branches & history
See all of it. The dashboard shows every memory, every fact about you, the inbox of what netti did, and how your memory is growing. The dashboard
Share exactly what you mean to. Share one page as a link, open a room for a team, publish a skill to the community. Sharing
Take it back. Forget one page, one fact, or a whole memory, in two deliberate steps. Forgetting

the shape of it#

Each project is a Context Map: dimensions (markdown pages like vision or technical), nodes (structured entities: people, decisions, artifacts), and a commit log (append-only history of every change). Branches fork the whole map; merges bring direction back to trunk.

Each project also carries a capture protocol: instructions telling the AI what to save and how, tuned to this project. A logistics project and a screenplay don't get the same advice.

A dimension describes the present. When something is superseded, netti replaces it in the body and records one dated line under a ## history heading at the end of that dimension. A session reading your memory never sees two versions of the truth side by side, and "why did we change this?" still has an answer. How it all fits together

every page#

start here

getting startedzero to your first context map in two minutes.how it worksthe whole loop in one page: capture, fold, retrieve, and the person in the middle of it.questions & answersthe short answers — what innernet is, how it connects, and how it compares.

your innernet

the dashboardthe crumb, the menu, your spaces, the view chips, the ground: how the place is laid out and how to move in it.projectsinside a memory: its pages, reading and editing them, tasks, living documents, handoffs, and its history.the you spaceyour day read back to you, the mirror, and the private ledger: every fact innernet holds about you, and who can see it.the inbox & the foldwhat netti did while you were away, how to answer it, and how a memory grows: folds and cells.sharingone page as a link, a fact as a card, a skill to the community. share exactly what you mean to.the chat windowinnernet inside your ai chat: one live line that opens into a ledger, and how to set it the way you like.shortcutsevery key and gesture the dashboard answers to, on one page.

for developers

for developersmemory that versions with your code. Connect once — it just works.repo memoryevery commit captured, every pull folded, one memory per repo, inherited on clone.

concepts

the context mapdimensions, nodes, commits, branches — and why memory is structured this way.retrievalhow one question becomes the few sections that answer it, inside a hard budget.branches & historyfork the whole memory, try a direction, diff it three-way, merge what won. and the one rule that keeps trunk readable.capture protocolper-project instructions that tell the AI what to save and what to skip.personal memorythe self map: what innernet knows about you, how every fact is classed, and what stays inside innernet.forgettingone page, one cell, one fact or a whole memory, gone in two deliberate steps.

connect an AI

connect an AIone URL, OAuth sign-in, done — wiring for every MCP client.the npx commandone command signs you in, wires every AI tool on the machine, gives the folder its own memory, and says what to try next.

build on the API

REST APIone memory, two doors — the REST API at full parity with the MCP tools.TypeScript SDKthe typed client — and why you may not need one.

agents

agentsmemory for the agents you keep: each one has its own, and what it learns about you comes to you.the agents apimake an agent, give it a memory, recall from it, and forget, over plain https.agent settingsevery setting an agent has, its default, and what it changes.

reference

MCP toolsevery hosted tool — and the REST endpoint each one mirrors.privacy & securityrls everywhere, hashed credentials, optional encryption at rest, full export/delete.glossaryevery word innernet uses, defined once, with where to read more.