Documentation

Quick Start

npm install -g notoken
notoken                    # interactive mode
notoken doctor             # diagnose system
notoken install claude     # install tools
notoken fix npm            # fix issues

Subcommands

notokenInteractive chat mode (default)

notoken doctorFull diagnostic + auto-fix

notoken install <tool>Install Claude, OpenClaw, Ollama, Docker...

notoken fix <target>Fix npm, git, docker, permissions, PATH, DNS

notoken setup <env>Set up dev/server/docker/node environment

notoken logs <service>Tail service logs

notoken heal:claudeClaude-powered auto-learning

notoken "<command>"One-shot NLP command

Flags

--dry-runParse but don't execute

--verboseShow detailed restatement

--explainShow full decision chain (why, alternatives, risks)

--jsonOutput as JSON

--yes, -ySkip confirmations

--auto-learnLearn from failures via LLM

Interactive Commands

<text>Execute natural language command

<text> &Run in background

:jobsBackground tasks

:play <name> [env]Run a playbook

:contextKnowledge tree

:historyConversation log

:platformOS/distro info

:backupsAuto-backup list

:auto-learnToggle auto-learning

:explainToggle explain mode

:helpFull command list

Ctrl+C twiceQuit and save conversation

Configuration

All config-driven. Add new commands by editing JSON — no code changes needed.

config/intents.json119 intent definitions

config/rules.jsonEnvironment + service aliases

config/hosts.jsonSSH targets per environment

config/file-hints.jsonKnown file locations by service

config/playbooks.jsonMulti-step playbook recipes

LLM Configuration

# Claude CLI (recommended)
export MYCLI_LLM_CLI=claude

# API endpoint (Claude or OpenAI)
export MYCLI_LLM_ENDPOINT=https://api.anthropic.com/v1/messages
export MYCLI_LLM_API_KEY=sk-ant-...

# Ollama (auto-detected, no config needed)
# Just run: ollama serve