Noisy Studio MCP server
Commit Nov 10, 2026Provides a voice interface for Claude Code, enabling speech recognition and spoken replies during agentic coding tasks.
Also known as noisy/noisy-coding
- Stars
- 14
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- 2.13.3
- Last commit
- Nov 10, 2026
- License
- MIT
What is Noisy Studio MCP?
Noisy Studio provides a voice interface for Claude Code, enabling speech recognition and spoken replies during agentic coding tasks.
Summary generated from the project's README
About Noisy Studio MCP server
Noisy Studio provides a voice interface for Claude Code, allowing developers to interact with their coding agents through speech recognition and spoken replies. This native macOS application for Apple Silicon includes a built-in Python engine, which removes the need for a separate service manager or manual Python installation. Users manage their experience through a desktop app that features a dashboard for tracking conversations and specific settings for voice and character profiles.
To use this server, developers must install the companion plugin via the Claude plugin marketplace using the noisy/noisy-studio package. The setup requires completing provider and audio configurations within the desktop application. While the plugin launches the engine bundled with the app, API credentials must be stored in the provider settings rather than in chat history. For those using offline providers, models are downloaded during the first use to enable local execution.
What you can do with it
- 1Voice-controlled agentic codingA developer can speak commands to their coding agent to guide it through complex programming tasks.
- 2Customized character interactionsYou can configure specific voice and character settings for every individual conversation.
- 3Monitoring active coding sessionsUsers can track and manage the various coding conversations they are currently following through a central dashboard.
Written from the project's README. Every use case is checked against a quote from it.
How to install Noisy Studio MCP
Formats verified 10 Nov 2026Install Noisy Studio MCP in Claude Code
$ claude mcp add --env XAI_API_KEY=<YOUR_XAI_API_KEY> noisy-coding -- docker run -i --rm docker.io/noisy/noisy-coding:2.13.3Install Noisy Studio MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"noisy-coding": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"docker.io/noisy/noisy-coding:2.13.3"
],
"env": {
"XAI_API_KEY": "<YOUR_XAI_API_KEY>"
}
}
}
}Install Noisy Studio MCP in VS Code
.vscode/mcp.json
{
"servers": {
"noisy-coding": {
"type": "stdio",
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"docker.io/noisy/noisy-coding:2.13.3"
],
"env": {
"XAI_API_KEY": "<YOUR_XAI_API_KEY>"
}
}
}
}Install Noisy Studio MCP in Claude Desktop
claude_desktop_config.json
{
"mcpServers": {
"noisy-coding": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"docker.io/noisy/noisy-coding:2.13.3"
],
"env": {
"XAI_API_KEY": "<YOUR_XAI_API_KEY>"
}
}
}
}Install Noisy Studio MCP in Codex
~/.codex/config.toml
[mcp_servers.noisy-coding] command = "docker" args = ["run", "-i", "--rm", "docker.io/noisy/noisy-coding:2.13.3"] [mcp_servers.noisy-coding.env] XAI_API_KEY = "<YOUR_XAI_API_KEY>"
Install Noisy Studio MCP in Gemini CLI
~/.gemini/settings.json
{
"mcpServers": {
"noisy-coding": {
"command": "docker",
"args": [
"run",
"-i",
"--rm",
"docker.io/noisy/noisy-coding:2.13.3"
],
"env": {
"XAI_API_KEY": "<YOUR_XAI_API_KEY>"
}
}
}
}Configuration
XAI_API_KEYSecretxAI API key for Grok voice synthesis and transcription (configured once on the dashboard, stored in the config volume)
| Name | Description | |
|---|---|---|
| XAI_API_KEY | xAI API key for Grok voice synthesis and transcription (configured once on the dashboard, stored in the config volume) | Secret |
Automated checks
Ran Nov 12, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work
- Source repository linked and reachablePass
- Open source license (OSI)Pass
- Commit or release in the last yearPass
- Package name not close to a popular packageNot checked
- Remote endpoint uses HTTPSNot checked
- Destructive tools are labelledNot checked
- No hidden instructions to the modelPass
- No critical advisory in the package or its dependenciesNot checked
README
View on GitHubNoisy Studio
Talk to your coding agent while it works. Noisy Studio is a native desktop app with speech recognition, spoken replies, per-conversation voice and character settings, and a dashboard for the conversations you are following.
https://github.com/user-attachments/assets/331ca147-b325-4b89-9704-a3bba206f2f8
Install
Download the macOS Apple Silicon app from Releases, move Noisy Studio.app to Applications, and open it. Complete provider and audio setup in the app. The app includes its own Python engine; ordinary Claude installation needs neither Python nor a separate service manager.
For Claude Code, install the companion plugin:
claude plugin marketplace add noisy/noisy-studio
claude plugin install noisy-studio@noisy
Restart Claude Code and ask it to set up Noisy Studio voice. Keep the app and
plugin versions together: the plugin launches the engine bundled with the app.
The internal plugin name remains noisy-studio for compatibility.
Installation and troubleshooting covers permissions,
providers, updates, and a spoken round trip. The Codex preview
uses the same app and currently requires uv for its integration processes.
macOS is the validated release platform; other platforms are source development.
Development
Use local development for the isolated daemon on 7765 and dashboard hot reload. The installed app owns 9765 and has separate settings. See desktop packaging for signed app builds.
- Agent hooks
- Ports
- Product naming
- Storybook - how the dashboard and the
widget are meant to look, published from
main. Run it locally withcd dashboard && npm run storybook. Its title shows the commit it was built from, so a story can be cited by URL and version.
API credentials belong in the app's provider settings, never in chat or source. Offline providers download their models on first use and then run locally.
Upgrading to the V3 technical names
V3 uses noisy-studio throughout packages, commands and integrations. Older installations need an explicit upgrade; see the agent-friendly upgrade and rollback guide before replacing an existing installation.
Noisy Studio MCP questions
How do I add Noisy Studio MCP to Claude Code?
Run claude mcp add --env XAI_API_KEY=<YOUR_XAI_API_KEY> noisy-coding -- docker run -i --rm docker.io/noisy/noisy-coding:2.13.3 in your terminal. Add --scope user to make it available in every project.
How do I add Noisy Studio MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
How do I install Noisy Studio for Claude Code?
Download the macOS Apple Silicon app from the Releases page, move it to Applications, and then install the companion plugin via the Claude plugin marketplace.
Does Noisy Studio require Python to be installed separately?
No, the app includes its own Python engine, so a standard Claude installation does not require Python or a separate service manager.
Can I use Noisy Studio offline?
Yes, offline providers will download their models on the first use and then run locally.
What platform is officially supported?
macOS is the validated release platform, while other platforms are currently for source development.