v3.1.6 · auto-updates enabled

Download c9ai

A local AI workbench — wiki-first, pure-black, runs on your own machine. Your notes, your chat, your action chips; all on disk, all yours. Install once; updates arrive in the background.

Windows

v3.1.6 · Windows 10 or later · x64

Download installer Portable (no install)

Signed installer · SmartScreen may prompt on first run (we're working on a cert).

Linux

Binary pending · install from source

npm run electron

Same Electron app, runs from a git clone. See Install from source below for the one-liner and deps.

macOS

Signed binary pending · install from source

npm run electron

Same Electron app, runs from a git clone. Signed .dmg arrives once the Developer ID is wired up.

Auto-update enabled. You install once. Every launch checks matsyaai.com/dist/c9ai/ for a newer build and downloads it in the background. When it's ready, you'll see a prompt to restart. Manual reinstall is never required after the first time.

First run

1
Install & open

Run the installer (Windows) or chmod +x the AppImage (Linux) and launch. A native window opens with the c9ai wiki — no terminal needed.

2
Add an AI provider

Press : in the wiki to open the command bar, run :settings, and paste an API key. Claude, OpenAI, Gemini and local Ollama are supported — any one is enough.

3
Start writing

Press n to create a page, :c9 to stream AI into it at your cursor, :chip to drop an action chip that runs a skill. The Timeline tab shows your feed of work — and your cohort's, if you subscribe to each other.

4
Connect to Matsya (optional)

On a Matsya workspace? Press : and run :matsya-api. A one-field form takes your MATSYA_API_KEY; c9ai saves it and starts polling the BRU queue to work tasks from the cloud on your behalf.

Install from source (Linux & macOS)

Same Electron app, same auto-update, pulled straight from GitHub. Takes about three minutes on a warm npm cache.

git clone https://github.com/hebbarp/c9ai-v3.git cd c9ai-v3/c9ai npm install npm run electron
Dependencies

Required (all platforms)

Linux only

  • libnss3, libgbm1, libxkbcommon0 — Electron runtime libs
  • On Ubuntu/Debian: sudo apt install -y libnss3 libgbm1 libxkbcommon0
  • On Fedora: sudo dnf install -y nss mesa-libgbm libxkbcommon

At least one AI provider key (Claude / OpenAI / Gemini) or Ollama running locally. Configure with :settings inside c9ai once it's running.

Wiki-first

Pages are the artifact. Chat writes into them; action chips turn words into verbs your alter-ego can run.

Live code blocks

Fenced code blocks with a c9ai:skill marker become callable skills on save. Literate programming, hot-swapped.

Social feed

Subscribe to a cohort-mate's c9ai. Their timeline merges into yours — a feed of making, not reactions.

Prerequisites

Required
  • At least one AI provider key (Claude, OpenAI, Gemini) or a local Ollama model
Optional
  • Matsya API key — to poll the BRU queue from your workspace
  • Ollama for offline local LLM (no API costs)