Log inSubmit server
Windy Word logo

Windy Word

Controls the Windy Word desktop app through MCP for transcription, paste workflows, hotkeys, models, installs, and local archive operations.

InstallRepositoryWebsite
Full README Compare alternatives Embed widget Claim this listing

Install Windy Word

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "windy-word": {
      "command": "npx",
      "args": [
        "-y",
        "windy-word-mcp"
      ],
      "env": {
        "WINDY_WORD_MCP_HOST": "YOUR_VALUE_HERE",
        "WINDY_WORD_MCP_PORT": "YOUR_VALUE_HERE",
        "WINDY_WORD_MCP_TIMEOUT_MS": "YOUR_VALUE_HERE"
      }
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Values are written into the snippet above so you can paste it as is. They never leave your browser.

Install not yet sandbox-tested— the server itself is confirmed live (we called its tools/list endpoint). The stdio install command is a separate, ongoing check.

Overview

Windy Word MCP server exposes Windy Word’s local voice-to-text desktop app to MCP agents through a schema-validated wrapper. It forwards tool calls to Windy Word’s localhost HTTP control server, enabling agents to manage transcription models, recording, paste strategies, hotkeys, dependencies, audio files, and archive exports. Reach for it when an agent needs to operate Windy Word on the user’s machine or diagnose platform-specific paste and input behavior.

Use cases

  • Control Windy Word recording and transcript pasting from an MCP agent
  • Diagnose Wayland, XWayland, hotkey, and paste-strategy issues
  • Transcribe individual audio files with the Windy Word Whisper engine
  • Switch Whisper models and inspect WindyTune performance state
  • Export archives and manage local voice-clone media

Key features

  • Local MCP wrapper for Windy Word’s HTTP control server
  • Whisper model selection and audio-file transcription
  • Platform-aware paste strategies and diagnostics
  • Global hotkey management
  • Whitelisted dependency installation with audit history
  • Archive, voice-clone, and sound-effect controls

Tools115

What Windy Word lets your agent do. Expand a tool for its parameters.

Verified live~14.8k tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

9/9 checks healthy over the last 57d

What it can do

From the 115 tools this server returned when we ran it.

Runs commands · 3Can delete or modify data · 8Handles credentials · 4Makes network requests · 6Read-only tools · 104
Network at startup3 connections
Files written at startupNone
Install scriptsNone
npm provenanceNot published
Package age149 days
Maintainers1

Observed 2026-10-17 in an isolated sandbox. A description of reach, not a security audit.

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 2d ago via OSV.dev · windy-word-mcp (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars1Stargazers on the source repository.
npm downloads233Package downloads in the last 30 days.
Last commit13d agoMost recent push to the default branch.
Tools exposed115Callable tools this server registers over MCP.
Directory activity3 viewsConfig copies, upvotes, and views on AllMCPs.

How Windy Word works

What Windy Word does

Windy Word MCP server connects MCP clients to the Windy Word desktop application, a local voice-to-text tool. The wrapper communicates with Windy Word over its HTTP control server on the local machine rather than sending state queries through a remote service. Windy Word must be running for the control connection to work, and the package requires Node.js 18 or newer.

The exposed interface covers several parts of the desktop app. Agents can start and stop recording, paste the latest transcript, choose a Whisper model, inspect microphones, change language settings, and query recording state. They can also transcribe an individual audio file through the same Python Whisper engine used for live transcripts. Supported input formats depend on what ffmpeg can read, with limits of 500 MB per file, 60 seconds for ffmpeg processing, and 120 seconds for the WebSocket request.

The Windy Word MCP server also provides controls for paste behavior, keyboard shortcuts, dependency installation, archive inspection, voice-clone workflows, and sound-effect settings. It belongs to the Speech-to-Text category, but its tools extend into local desktop automation and app administration.

How it works

An MCP client launches the Node-based server, which validates tool requests and forwards them to Windy Word’s local control endpoint. The default endpoint is 127.0.0.1:18765. Host, port, and request timeout values can be overridden with environment variables.

Paste diagnostics account for operating-system differences. Agents can identify the display server, desktop environment, focused-window type, available tools, and the selected fallback chain. The server supports strategy selection across macOS, Windows, Linux X11, and Linux Wayland. Some tools inject text into the currently focused window, so a user should focus a safe target before running those tests. The end-to-end injection test instead creates a temporary Tk scratchpad and is limited to Wayland with GNOME in the documented version.

Model changes can hot-reload the running Python engine over WebSocket. WindyTune reports recent timing data and uses its configured model ladder to decide when to move between models.

Setup and configuration

Install the package through an MCP client with npx -y windy-word-mcp, or run the cloned repository with Node after installing its npm dependencies. A typical Claude configuration starts the package as a stdio server. Windy Word itself must be open so its local HTTP control server is available.

Supported configuration variables are WINDY_WORD_MCP_HOST, WINDY_WORD_MCP_PORT, and WINDY_WORD_MCP_TIMEOUT_MS. The timeout is a default for requests; installation and transcription operations apply their own longer-operation handling.

Linux dependency installation is restricted to an allowlist and may use pkexec. The user must approve the graphical authentication prompt unless the documented polkit setup has been enabled. Installation history and asynchronous job state are held in memory and reset when the application restarts.

Tools and capabilities

The Windy Word MCP server includes tools for:

  • Recording, transcript pasting, Whisper model selection, language configuration, and audio-file transcription.
  • Paste-strategy selection, fallback testing, target detection, paste history, and hotkey rebinding.
  • Platform diagnostics, dependency discovery, synchronous or asynchronous installation, and install auditing.
  • Archive listing, searching, date filtering, deletion, statistics, and soul-file zip export.
  • Voice-clone management, clone submission, media scanning, and folder-based ingest workflows.
  • Sound-hook configuration, sound-pack selection, master volume, and sound-effect modes.

Archive and installation histories are in-memory where documented. Soul-file export writes a zip containing the user’s audio, video, transcripts, and a manifest to a requested path.

Limitations and notes

The application must be running locally, and operations act on the user’s machine. Paste tests can inject text into the focused window. Sound-effect state is currently read from renderer localStorage; live reads may report that the renderer is not readable until a later renderer-side IPC bridge is available. Linux-only behavior includes the polkit setup flow and the documented GNOME focus-policy injection test. Bulk audio processing is not a single transcribe operation; agents should iterate over files and call the audio transcription tool for each one.

Read the full README View source on GitHub

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about Windy Word

Windy Word is an MCP server that connects MCP clients to the local Windy Word voice-to-text desktop app. Its main tools control recording, transcript pasting, Whisper models, hotkeys, paste strategies, audio-file transcription, dependency installation, archives, voice clones, and sound settings through Windy Word’s localhost control server.

Install it with `npx -y windy-word-mcp` through an MCP client, then make sure Windy Word is running locally. The package requires Node.js 18 or newer. A cloned repository can also be run with Node after `npm install`.

It supports `WINDY_WORD_MCP_HOST`, `WINDY_WORD_MCP_PORT`, and `WINDY_WORD_MCP_TIMEOUT_MS`. They override the control-server host, port, and default request timeout; the defaults are `127.0.0.1`, `18765`, and `5000` milliseconds.

Yes, it can inspect platform details, detect the focused target type, list available paste strategies, identify hotkey collisions, and test fallback chains. The end-to-end scratchpad injection test is documented for Wayland with GNOME.

Yes, on Linux it can list and install whitelisted tools such as `wtype`, `ydotool`, `wl-clipboard`, and `xdotool`. Installation may show a graphical polkit authentication prompt, while the optional polkit setup can enable prompt-free later installs.

Related MCP servers

All in Speech-to-Text All alternatives
Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewWindy Word AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/windy-word?style=directory)](https://allmcps.com/mcp/windy-word)
HTML Embed
<a href="https://allmcps.com/mcp/windy-word"><img src="https://allmcps.com/api/badge/windy-word?style=directory" alt="Windy Word on AllMCPs" /></a>