Nextbase CLI

Wisper is the first tool in Nextbase CLI: a voice-to-text utility that remains available as the wisper command. NoteBot is the second tool: it records meetings and turns them into transcripts, summaries, decisions, and action items. Future Nextbase command-line tools will live in this repository alongside them.

Nextbase umbrella command

Use the umbrella command when you want a menu:

nextbase

It shows:

1. Wisper  - dictation / polish / spell fix
2. NoteBot - meetings / audio notes / tasks

You can also route directly through it:

nextbase wisper setup
nextbase notebot open

Individual tool commands still work:

wisper setup
notebot open

macOS listener behavior

wisper listen starts the listener in the background so it survives closing Terminal. Use foreground mode only for debugging:

wisper listen --foreground

Windows and macOS also support modifier-only shortcuts like Ctrl+Window / Ctrl+Command when typed directly:

wisper shortcut Ctrl+Window

For login startup on macOS, use:

wisper autostart on

This installs a LaunchAgent and avoids Terminal-owned listener processes.

NoteBot — meeting notes

notebot setup
notebot meeting start
# When the meeting ends:
notebot meeting stop
notebot audio ./meeting.wav
notebot audio https://example.com/meeting.mp3
notebot open
notebot history
notebot tasks

notebot setup asks for missing keys locally:

• Sarvam or Groq key for multilingual transcription
• Groq key for meeting summaries, decisions, and action items

Meeting notes are stored locally under ~/.notebot/. Responsibilities are assigned only when explicit in the transcript; otherwise they are marked suggested or unassigned. notebot open launches a local dashboard with Start Meeting, Stop & Generate Notes, meeting history, transcript, decisions, tasks, a Remote URL input, and a Choose Local File button for uploading audio directly from the browser.

Install without Git

Windows PowerShell

iwr -useb https://raw.githubusercontent.com/Nextbasedev/nextbase-cli/master/install.ps1 | iex

The installer also tries to install SoX automatically with winget if it is missing.

Then:

nextbase

macOS / Linux

curl -fsSL https://raw.githubusercontent.com/Nextbasedev/nextbase-cli/master/install.sh | bash

Then:

nextbase

To update later without redoing setup:

wisper update

The updater installs the latest version, keeps existing API key/shortcut/history, asks only for missing new options, refreshes autostart, and restarts the listener.

The installer downloads the repo, builds it, and links nextbase, wisper, and notebot into your user bin directory.

Clean CLI-first base for a Wispr Flow-style dictation tool.

Simple UX

Start with one command:

wisper setup

It asks you to choose model from a menu, paste API key, verifies it, captures shortcut by pressing keys, asks whether to enable startup automatically, then starts the listener immediately.

Useful commands:

wisper provider   # choose provider from menu + verify key
wisper polish on  # enable auto polish before paste
wisper polish shortcut # set selected-text polish shortcut
wisper spell shortcut # set focused-input spelling-fix shortcut
wisper spell shortcut CommandOrControl+Alt+S # set it directly
wisper polish "rough dictated text" # polish text manually
wisper media on 35 # lower system volume to 35% while recording
wisper media test # test ducking/restoring system volume
wisper autostart on # enable startup listener
wisper autostart off # disable startup listener
wisper autoupdate status # check background auto-update setting
wisper autoupdate check # check GitHub for new version
wisper autoupdate check --apply # update now if available
wisper shortcut F15 # set dictation shortcut directly
wisper shortcuts # show supported shortcut keys
wisper mic --auto # test microphones and pick working one
wisper status     # show current setup
wisper update     # install latest and only ask missing setup prompts
wisper listen     # run background listener
wisper stop       # stop background listener
wisper restart    # restart background listener
wisper logs       # show listener logs
wisper open       # open local web app

Base features

• CLI entrypoints: nextbase, wisper, and notebot
• Local transcript storage in ~/.wisper-cli/history.json
• wisper setup for simple first-time setup
• wisper provider to choose provider from a menu
• wisper polish on/off to enable or disable auto-polish before paste
• wisper polish shortcut to set the global selected-text polish shortcut
• wisper spell shortcut to fix spelling in the entire focused text input without manual selection
• wisper polish "text" to rewrite text manually with Groq polish mode
• wisper media on/off/status/volume/test to control Windows audio ducking while recording
• wisper autostart on/off/status to control startup listener without rerunning setup
• wisper autoupdate on/off/status/check to keep the background listener updated automatically
• wisper shortcut [key] to set shortcut from a prompt or directly, e.g. wisper shortcut F15
• wisper shortcuts to show current shortcuts and supported keys
• wisper mic --auto to record tiny test samples and pick the working microphone
• Windows listener watches for newly connected/removed microphones and auto-switches to the strongest working input
• wisper status to show current setup
• wisper update to install latest version while preserving setup
• wisper listen background listener target
• automatic startup after wisper setup
• optional auto-polish mode powered by Groq chat completions (llama-3.3-70b-versatile)
• selected-text polish shortcut: select text anywhere, press CommandOrControl+Shift+P, and Wisper replaces it with polished text
• direct F-key setup for terminals that cannot capture F13-F24: wisper shortcut F15, wisper polish shortcut F16
• focused-input spell fix: focus an editable field and press CommandOrControl+Alt+S; Wisper selects all, fixes spelling only, and replaces the field content
• background auto-update: listener checks GitHub periodically, installs new builds silently, and restarts itself
• optional Windows audio ducking: lower system/media volume while holding shortcut, then restore after release
• wisper history to print transcript history
• wisper add "text" to save a manual transcript while the base is being built
• wisper app / wisper open to launch a local web dashboard
• Local web dashboard at http://127.0.0.1:3838

Not included in this base

• Meeting transcription
• Desktop/Tauri app shell
• Full recorder/hotkey implementation
• Cloud sync

See docs/PLAN.md for the full completion plan.

Planned next features

1. wisper record — record mic audio from CLI.
2. wisper transcribe <file> — transcribe audio file.
3. Provider adapters — Groq first, then ElevenLabs/Sarvam.
4. wisper polish "text" — rewrite dictated text.
5. Settings command + local config file.
6. Better history search/filter/copy in the web dashboard.