What is it?
What it is
A Claude Code installer wizard that assembles a full-stack AI agent with persistent memory, voice, visual face, and optional hand-controlled interface.
Why it exists
To provide an easy guided setup that combines four independent open-source projects into a single conversational agent, so users can quickly get a living circuit board with voice and memory.
Who should use it
Developers and power users with a Claude Code subscription (Pro $20+) who want a guided installer that wires together a persistent AI memory vault, voice interface, animated visualizer face, and optional webcam-based hand gesture control into a single personal AI agent. Especially valuable for users who have hand-built partial setups with older prompts and want to consolidate or upgrade them, and for Mac/Linux/Windows users who want an install wizard that handles dependencies (git, Python) automatically.
Who should avoid it
Users without a Claude Code subscription or who do not want to rely on Claude Code as the runtime, anyone looking for a traditional full-stack application framework (this is an installer/orchestration repo, not a web/backend stack), users unwilling to grant the wizard access to their home directory or run a guided installation, and those who need a closed-source commercial license without contacting the author for a separate agreement.
How it works
A quick walkthrough in plain English
How fullstack-agent works
Step 1 of 3
You interact with it
Open fullstack-agent, send a request, or connect it to your stack.
Features
Advantages
- Remembers user and work across sessions
- Provides natural spoken interaction
- Full-screen visualizers sync with conversation
- Optional hand gestures for screen control
- Free and open source
- Commercial use allowed
- Active community and support
- Easy updates via git pull
- Customizable agent identity
Disadvantages
- Requires Claude Code and subscription
- Voice needs microphone and ~1 GB local models
- Hands need webcam and Chrome
- Windows requires extra setup for Python
- Installation requires git (except Windows zip method)
- May need to leave window open during install
- Potential cross-piece configuration issues
Installation
native
On Mac and Linux: create a directory (~/my-agent), clone the repository with 'git clone https://github.com/jaredrhod/fullstack-agent', navigate to the fullstack-agent folder, and run 'claude "set me up"' in Claude Code. On Windows: use the provided PowerShell script to download the toolbox as a zip, expand it, copy the contents to the fullstack-agent directory, and then run 'claude "set me up"' in Claude Code.
FAQ
What does fullstack-agent give me?
It assembles four pieces into one living AI agent: a persistent memory vault (ai-memory-vault), a voice system (backtalk) that answers aloud, a full‑screen visualizer (ai-visualizer) that shows a face synced to conversation, and an optional hands module (barehands) that lets you move notes and images with your webcam. You can choose any subset of these components during setup.
How do I install fullstack-agent?
On macOS/Linux, clone the repo into ~/my-agent/fullstack-agent and run `claude "set me up"`. On Windows PowerShell, run the provided script that downloads the repo as a zip, extracts it, and then runs the same wizard. You need a Claude Code subscription (the $20 Pro plan is enough) and git (the installer sets it up on Windows).
Can I use only some of the pieces (memory, voice, face, hands)?
Yes. Each piece is optional. The wizard asks which components you want, explains what each does in plain English, and installs only the ones you select. You can add missing pieces later by running the wizard again.
How do I launch my agent after setup?
The wizard leaves three shortcuts on your Desktop named after your agent: **Chat** (typed session, terminal only), **Talk** (voice + face), and **Barehands** (voice + hands board). Double‑click the mode you want; you can also run `fullstack-agent/start.sh` (or `start.bat` on Windows) from the terminal.
How do I update everything?
Run `./fullstack-agent/update.sh` on macOS/Linux or `fullstack-agent\update.bat` on Windows. Updates pull the latest changes from each repo while preserving your agent’s identity and memory vault, which live outside the repos.
What if something breaks or is confusing?
Ask your agent to fix it – it has built‑in troubleshooting guides and is instructed to resolve issues automatically. Refer also to the `TROUBLESHOOTING.md` file here and each component’s own README for detailed help.
Featured in Videos
YouTube tutorials and walkthroughs for fullstack-agent
Alternatives
Similar projects ranked by category, topics, and text overlap.
