Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By harshav167
Voice conversations with AI coding assistants using ElevenLabs TTS/STT. HTTP server on port 8765 with local Silero VAD and manual commit mode.
npx claudepluginhub harshav167/ava --plugin avaRemote voice via Ava Connect. Use when users want to add voice to an AI coding assistant using their phone or web app, without local STT/TTS setup.
Background music control for Ava voice sessions using mpv
Voice interaction for AI coding assistants. Provides natural voice conversations using ElevenLabs TTS and STT. Use when users mention ava, speak, talk, converse, voice status, or voice troubleshooting. ElevenLabs-only: eleven_v3 TTS model, Scribe v2 Realtime STT with local Silero VAD.
Restart the Ava dev server and verify it's healthy. Use after code changes.
Run tests and import checks to verify the codebase is healthy. Use after making changes or before committing.
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Voice conversations with Claude Code using local speech-to-text and text-to-speech
Claude Code skill pack for ElevenLabs (18 skills)
Audio feedback when Claude Code agent completes tasks using pocket-tts
Voice I/O for Claude Code - speak to users and hear their voice input
Voice-first interaction with automatic TTS markers for Claude Code
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Claude Code plugin for Plaud AI — manage recordings, transcripts, summaries, and audio via CLI commands, MCP tools, prompts, and skill resources
Voice conversations for AI coding assistants over MCP.
Ava runs a local MCP HTTP server, speaks through ElevenLabs TTS, listens through ElevenLabs Scribe realtime STT, and coordinates the microphone with a conch lock so multiple agents do not talk over each other.
eleven_v3 TTS with configurable voice, model, speed, and instructions127.0.0.1:8765Install dependencies and start the local server:
uv sync
./scripts/ava-server.sh setup
./scripts/ava-server.sh status
Add the MCP server to a client:
{
"mcpServers": {
"ava": {
"type": "http",
"url": "http://127.0.0.1:8765/mcp"
}
}
}
Required user config lives in ~/.ava/ava.env:
ELEVENLABS_API_KEY=sk_...
AVA_ELEVENLABS_TTS_MODEL=eleven_v3
AVA_ELEVENLABS_STT_MODEL=scribe_v2_realtime
AVA_TTS_SPEED=1.2
Project overrides can go in a repo-local .ava.env.
uv run ava --version
uv run ava serve --host 127.0.0.1 --port 8765
uv run ava converse
uv run ava claude hooks list -s all
uv run ava soundfonts status
The installed command is ava. The Python package is ava.
The repo includes a launchd wrapper for local development:
./scripts/ava-server.sh setup
./scripts/ava-server.sh start
./scripts/ava-server.sh stop
./scripts/ava-server.sh restart
./scripts/ava-server.sh status
./scripts/ava-server.sh logs
./scripts/ava-server.sh health
Logs are written under ~/.ava/logs/. The main structured log is ~/.ava/logs/ava.log; launchd stderr is ~/.ava/logs/server.err.
After code changes, restart the server when you are ready for the running process to pick them up. Do not restart during active playback or a live voice turn.
Hooks are opt-in. They are not installed by default.
uv run ava claude hooks add
uv run ava claude hooks list -s all
uv run ava claude hooks remove
Installed hooks can play soundfont feedback for tool events and support Ava Connect session/notification behavior. Soundfont playback defaults to enabled only after hooks are installed; disable it with:
uv run ava soundfonts off
uv run ava soundfonts off --config
Use uv for Python work:
uv run pytest -q
uv build
cd installer && uv build
Useful focused checks:
uv run python -c "from ava.server import mcp; from ava import __version__; print(__version__)"
uv run ava --version
ava/ - Python package and MCP server implementationava/tools/ - MCP tool implementationsava/data/hooks/ - Claude Code hook definitions and shell receiversava/templates/ - launchd/systemd/script templatesinstaller/ava_install/ - installer packagescripts/ava-server.sh - local launchd server helperdocs/ - user and developer documentationtests/ - pytest suiteThis project is thanks to the great work done by the original upstream repo, mbailey/ava, created by Mike Bailey and the Failmode project. This fork keeps that foundation and adapts the codebase around the renamed Ava package, ElevenLabs-first runtime, local MCP server workflow, and repo-specific development setup.
MIT. See LICENSE.
mcp-name: com.failmode/ava