Release v4.4.0
Added
-
MCP Registry Support
- Add server.json configuration for MCP registry publication
- Add mcp-name field to README for PyPI package validation
- Integrate MCP registry publishing into CI/CD workflow
- Support DNS-based namespace authentication (com.failmode/voicemode)
- Update Makefile to sync server.json version during releases
-
Cloudflare Worker for voicemode.sh
- Serve install script via custom domain
- Smart user-agent detection for CLI vs browser
- Cached script delivery with fallback
-
Selective Tool Loading
- Reduce token usage by loading tools on demand
- Implement smart tool filtering based on context
- Add tool loading configuration options
-
Documentation Improvements
- Complete documentation reorganization
- Add tutorials, guides, and reference sections
- Improve getting-started guide with clear paths
- Add universal installer as primary quick start
- Archive outdated documentation
-
Three Bears Agent Support
- Add baby-bear, mama-bear, and papa-bear agent configurations
- Integrate with sound fonts for agent-specific audio feedback
Changed
- Consolidate PyPI and MCP Registry publishing workflows
- Update branch references from 'main' to 'master'
- Improve Cloudflare Worker error handling and caching
- Rename hook to hooks, stdin-receiver to receiver
Fixed
- Fix broken documentation links after refactor
- Restore minimal claude command group for hook support
- Fix Claude settings.json path configuration
Installation
Universal Installer (Recommended)
Install Claude Code with Voice Mode configured and ready to run on Linux, macOS, and Windows WSL:
curl -O https://getvoicemode.com/install.sh && bash install.sh
This installer will:
- Install all system dependencies (Node.js, audio libraries, etc.)
- Install Claude Code if not already installed
- Configure Voice Mode as an MCP server
- Set up your system for voice conversations
Manual Installation
PyPI
pip install voice-mode==4.4.0
Claude Code MCP
claude mcp add voice-mode uvx voice-mode