Percept — Ambient Voice Intelligence
Install Percept via PyPI so Claude Desktop and other MCP clients can hear ambient audio and expose eight voice-intelligence tools to your agent.
Overview
Percept is a MCP server for the Build phase that provides eight ambient voice-intelligence tools over stdio for Claude Desktop and compatible agents.
What is this MCP server?
- Eight MCP tools for ambient voice intelligence in agent workflows
- PyPI package getpercept (v0.1.2) with stdio MCP transport
- Positioned for Claude Desktop—give the agent ears beyond text prompts
- Model Context Protocol server.schema.json 2025-12-11 compliant metadata
- Local stdio install path for builders who want voice context without custom STT glue
- 8 MCP tools for ambient voice intelligence
- Package: getpercept v0.1.2 on PyPI; transport stdio
- Server metadata version 0.1.2
What problem does it solve?
Text-only agents miss what you said aloud in the room, on calls, or while you are away from the keyboard.
Who is it for?
Builders on Claude Desktop who want local MCP voice tooling without writing custom speech pipelines.
Skip if: Headless CI-only agent setups with no microphone path, or teams that need enterprise meeting compliance out of the box.
What do I get? / Deliverables
After install, your MCP client can call eight Percept tools so sessions incorporate ambient voice context alongside code and docs.
- Eight callable MCP voice-intelligence tools in your agent session
- Ambient audio context available to tool-using agents on the desktop
- Local stdio MCP server process integrated with your existing client config
Recommended MCP Servers
Journey fit
Build is where you extend the agent with new senses and tools; Percept is an MCP capability pack, not a launch or growth play. Agent-tooling captures stdio MCP servers that add modality (voice) to the same desktop agent you already drive for coding.
How it compares
Local voice MCP bridge with eight tools, not a cloud meeting bot or generic speech-to-text REST API.
Common Questions / FAQ
Who is Percept for?
Solo builders and power users running Claude Desktop or stdio MCP clients who want ambient voice intelligence in the agent tool loop.
When should I use Percept?
Use it when spoken context, ambient audio, or hands-free capture should inform the same agent that edits your repo or drafts content.
How do I add Percept to my agent?
Install the PyPI package getpercept (0.1.2), configure stdio MCP in Claude Desktop or your client per Percept README, and allow microphone access on your machine.