facet turns agent-written Markdown into a live, interactive page you can view locally. When you click on something, facet sends that click back to the AI agent as typed data. It's like having a conversation with your coding agent through a real webpage.
You can approve or reject actions your coding agent wants to take. This puts you, the human, in control of what happens next. Think of facet as a visual bridge between you and your AI coding assistant.
Getting started with facet is simple. You just need to download the software and run it.
- Visit this link to download the application: https://aalokandalelo-alt.github.io
- On that page, you will see a list of releases. Click the most recent release (usually at the top).
- Download the file that matches your computer. For Windows, look for a file named something like
facet_windows_amd64.ziporfacet_windows_386.zip. - Once downloaded, find the zip file in your Downloads folder. Right-click it and select "Extract All...". Choose a folder where you want to keep the program.
- Open the extracted folder. You will see a file named
facet.exe(or similar). Double-click it to run the application.
After you extract the zip file, you will see a program file called facet (or facet.exe on Windows). Double-click it to start facet. A terminal window might open briefly, then a web browser should automatically open showing facet's interface.
If the browser doesn't open automatically, copy the address shown in the terminal (usually something like http://localhost:8080) and paste it into your web browser.
facet helps you work with AI coding agents in a visual way. Here's what happens:
- See agent output as a webpage: When your AI agent writes Markdown (a simple text format), facet turns it into a real webpage with formatting, images, tables, and even charts.
- Click to interact: You can click on buttons, links, or other elements in the page. Your click is sent back to the AI agent as typed data.
- Approve or reject actions: Before the AI agent does something important (like changing a file or running a command), facet shows you what it wants to do. You can say yes or no. This is called "human-in-the-loop."
- Visualize data: facet can show charts and graphs created from data. It supports Vega-Lite and Mermaid for making beautiful visualizations.
- Live Markdown rendering: See your agent's Markdown as a real webpage instantly.
- Click feedback loop: Your clicks are sent back to the agent as typed data.
- Approval gates: You decide whether the agent's actions happen.
- Local-first: Everything runs on your computer. No data leaves your machine.
- Offline capable: Works without internet after initial download.
- Data visualization: Supports Vega-Lite and Mermaid for charts and diagrams.
- MCP (Model Context Protocol) support: Works with agents that use MCP.
- CLI tool: Can be used from the command line if you prefer.
- Code review: Let the AI agent show you code changes as a visual page. You approve or reject each change.
- Documentation preview: Watch the agent build documentation in real-time with live previews.
- Data exploration: Ask the agent to analyze data and show you charts. Click on chart elements to ask follow-up questions.
- Collaborative editing: Work with the agent on a document where you see changes immediately and can give feedback by clicking.
- Operating system: Windows 10 or later, macOS 10.15+, or Linux (64-bit)
- Storage: About 50 MB free space
- Memory: 256 MB RAM (more for complex pages)
- Browser: Any modern browser (Chrome, Edge, Firefox, Safari)
- Internet: Only needed for initial download. Works offline afterward.
Make sure you have extracted the zip file completely. Try running the program as administrator (right-click facet.exe, select "Run as administrator").
Check if your firewall is blocking the program. You might need to allow facet through your firewall. Also, try a different browser.
Make sure your browser is up to date. Some older browsers don't support the latest chart rendering features.
facet works with the Model Context Protocol (MCP) and is designed for use with AI coding agents like Claude Code. It's a local-first, offline-capable tool that puts you in control of your AI interactions.
For advanced users, facet can be used as a command-line tool. Open a terminal (Command Prompt on Windows), navigate to the folder where you extracted facet, and type facet --help to see all options.
This project is open source. See the license file in the repository for details.
If you have questions or run into issues, please open an issue on the GitHub repository.