DEDUPED Hot AI repos, turned into use cases Get the Monday 5

How to give a Claude Code voice agent a live face with AI Visualizer

Point the visualizer at a backtalk signal folder so the face follows the real conversation.

Short answer: Run git clone https://github.com/jaredrhod/ai-visualizer and cd ai-visualizer; then copy ai-visualizer.json.example to ai-visualizer.json, then set name to your agent's name and face to the face the root URL should open; set bus_dir in ai-visualizer.json to your backtalk folder, or set signals_dir in backtalk's config to this folder. All steps.

  • Content
  • Intermediate
  • 30 minutes
  • From the README
  • Claude Code

Remote script. The README suggests pasting a one-sentence prompt into an existing Claude Code session so the agent clones the repo and runs the setup wizard on its own, so review what the agent runs before you approve it.

Who it is forA developer running a voice agent through Claude Code with backtalk who wants the agent to have a visible face during live sessions.

The problemA voice agent running in the background gives no visual feedback about whether it is listening, thinking or speaking.

The outcomeThe chosen face idles, listens, thinks and speaks in sync with the real voice conversation, and the agent name appears on the chip and in every HUD.

What you need first

  • Python 3 (ships with macOS and most Linux systems; on Windows, use run.bat or python server.py)
  • A web browser
  • Optional: a voice line or agent that writes the three signal files, such as backtalk, for live use

Steps

  1. Run git clone https://github.com/jaredrhod/ai-visualizer and cd ai-visualizer
  2. Copy ai-visualizer.json.example to ai-visualizer.json, then set name to your agent's name and face to the face the root URL should open (inferred, check the README)
  3. Set bus_dir in ai-visualizer.json to your backtalk folder, or set signals_dir in backtalk's config to this folder
  4. Run ./run.sh, restart backtalk, then say something to your agent and watch the face respond

Prompt for your agent

In the ai-visualizer repo, copy ai-visualizer.json.example to ai-visualizer.json, set name to <agent name> and bus_dir to <backtalk folder>, then run ./run.sh. Expected output: the root URL opens the configured face. Verify that the .voice_state file in the bus folder changes between idle, listening, thinking and speaking during a spoken exchange.

Caveat. Live voice sync only works when a voice line such as backtalk writes the three signal files, and the README says the visualizer is designed to pair with backtalk out of the box.

Use this in your AI agent

Ask your agent

Using the Deduped MCP, get the setup steps for jaredrhod/ai-visualizer and "Give a Claude Code voice agent a live face"

Add Deduped to Claude Code once

claude mcp add --transport http deduped https://dedupedai.com/mcp

Cursor, ChatGPT, others

Share: X LinkedIn Hacker News Email Copy link

Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer