Getting Started
Give Your AI a Brain
3 commands. 2 minutes. Your AI coding assistant will never forget again.
What NORDON Does
NORDON is a local Rust daemon that runs on your machine. It hooks into Claude Code, captures what happens during your sessions (decisions, failures, patterns), scores each event using 18 ML modules, and stores the important ones as typed memories.
Then it writes the most valuable knowledge into your project's CLAUDE.md file. When Claude Code starts a new session, it reads CLAUDE.md and instantly knows your project's history. Your AI gets smarter with every session.
The Feedback Loop
This is the core concept — understand this and you understand NORDON:
1. You code with Claude Code (normal usage)
↓
2. NORDON hooks capture every tool use (automatic)
↓
3. Rust ML scores each event (decisions > typo fixes)
↓
4. High-scoring events → typed memories (Decision, Failure, Constraint...)
↓
5. Knowledge Graph builds connections
↓
6. Reasoning engine produces inferences
↓
7. CLAUDE.md auto-updated with top knowledge
↓
8. Next session → Claude Code reads CLAUDE.md → smarter
↓
9. Loop repeats → every session gets better
You don't need to do anything after install. The loop runs automatically.
Install
Three commands. That's it.
npm install -g @sodasoft/nordon-cli
nordon init # creates config + auto-installs Claude Code hooks
nordond & # starts the local ML daemonnordon init automatically installs Claude Code hooks in ~/.claude/settings.json. You never need to edit that file manually.Verify
$ nordon doctor
PASS Config file exists
PASS Data directory exists
PASS Database exists
PASS Daemon is running
PASS Hooks installed
All checks passed.$ nordon status
Daemon: running
Uptime: 2m
Memories: 0
Repos: 0
Events: 0What Happens Automatically
After install, NORDON does all of this without you doing anything:
- → Hooks capture every Claude Code tool use (Edit, Write, Bash, Read)
- → ML scores each event — decisions get ~0.85, failures ~0.70, boring reads ~0.30
- → Events above 0.4 become typed memories (Decision, Failure, Constraint, Pattern...)
- → Knowledge Graph auto-builds connections between related memories
- → Reasoning engine detects risks and chains rules
- → CLAUDE.md is auto-updated every 10 minutes with your top knowledge
- → Secrets are auto-redacted (31 patterns: API keys, tokens, passwords)
- → Sensitive files are auto-blocked (.env, credentials, SSH keys)
VS Code Extension
Install from VS Code Marketplace (search "NORDON"). The extension:
- → Auto-connects to daemon on launch
- → Auto-installs hooks if not already installed
- → Shows memory count in status bar
- → Knowledge Graph visualization (Command Palette → "NORDON: Knowledge Graph")
- → Sidebar with status, hooks, memories, brain inferences
Useful Commands
| Command | What it does |
|---|---|
| nordon status | Daemon status, memory count, health grade, top inference |
| nordon summary | Generate a project context snapshot |
| nordon memories list | All memories as formatted table |
| nordon memories search <q> | Search by keyword |
| nordon memories review | Interactive approve/reject flow |
| nordon playbook | Auto-generate onboarding doc |
| nordon graph | Show knowledge connections |
| nordon suggestions | Actionable improvement tips |
| nordon analytics | Usage statistics with charts |
| nordon sync-knowledge | Write knowledge to CLAUDE.md now |
| nordon quick-ref | Full command reference card |
That's it!
Use Claude Code normally. NORDON works in the background — capturing, scoring, storing, and injecting. After a few sessions, check CLAUDE.md — you'll see a "Project Knowledge" section auto-generated from your sessions.