Mcp Listen
Add microphone capture and speech-to-text to local agent workflows so coding assistants can hear voice input during builds.
Overview
mcp-listen is an MCP server for the Build phase that provides microphone capture and speech-to-text so AI agents can process voice input.
What is this MCP server?
- Microphone capture exposed to MCP clients on the developer machine
- Speech-to-text pipeline for turning spoken prompts into agent-usable text
- npm mcp-listen package with stdio transport (v0.1.3)
- Focused scope: listening input, not full voice assistant product stack
- npm package mcp-listen version 0.1.3
- Described capabilities: microphone capture and speech-to-text
Community signal: 5 GitHub stars.
What problem does it solve?
Coding agents are text-only by default, forcing builders to retype spoken ideas or maintain separate dictation apps outside MCP.
Who is it for?
Solo builders experimenting with voice-controlled dev agents or local STT-backed automations.
Skip if: Production call-center transcription or teams that cannot grant microphone access on developer machines.
What do I get? / Deliverables
After install, your agent can capture microphone audio and receive transcribed text inside the same MCP session as your code tools.
- Voice-to-text bridge usable from agent tool calls
- Hands-free prompt path into your coding session
- Reusable stdio MCP config for listen-capable workflows
Recommended MCP Servers
Journey fit
How it compares
Voice capture MCP module, not a full speech assistant SaaS or telephony integration.
Common Questions / FAQ
Who is mcp-listen for?
Indie builders and agent authors who want microphone and speech-to-text available to MCP clients without custom audio plumbing.
When should I use mcp-listen?
Use it during Build when you are adding voice input to an agent workflow or your personal coding assistant setup.
How do I add mcp-listen to my agent?
Install the mcp-listen npm package and register it as a stdio MCP server in Claude Code, Cursor, or your host configuration.