Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.q-uestionable.ai/llms.txt

Use this file to discover all available pages before exploring further.

The q-ai Web UI is a browser-based interface for running workflows, examining results, and managing provider credentials. It provides a visual alternative to CLI operations with real-time feedback and organized result viewing.

​
Launching the Web UI

Start the server with qai ui: 
qai ui
This launches the FastAPI backend and automatically opens your browser. The port is auto-selected (first available); the URL is printed to the terminal.

​
Custom Port

Specify a fixed port: 
qai ui --port 8080

​
Headless Mode

Start the server without opening a browser: 
qai ui --no-browser
The server will still be accessible at the default address. Use this when running on remote machines or in automation contexts.

​
Tech Stack

The Web UI is built with:
  • Backend: FastAPI with async request handling
  • Templates: Jinja2 with server-side rendering
  • Frontend: HTMX for dynamic updates without full page reloads
  • Real-time Updates: WebSocket connections for live workflow progress and result streaming
  • Data Storage: SQLite database for run history and configuration

​
Assistant Landing Page

When the assistant is configured, the default landing page (/) is the assistant chat interface. The workflow launcher moves to /launcher. If the assistant is not configured, the launcher remains the default landing page. See Assistant Overview for details.

​
Main Views

The Web UI has five primary sections:

​
Assistant

Chat with the built-in assistant to discover capabilities, interpret scan results, and plan testing workflows. The assistant also appears as a contextual panel on run results pages for asking questions about specific findings. Learn more about the Assistant

​
Launcher

Run new workflows interactively. Customize workflow parameters through accordion-based forms, preview the equivalent CLI command, and launch with a button click. Learn more about the Launcher

​
Runs

View workflow results after execution. Browse run history, examine module-specific findings, manage multiple concurrent runs, and export results in multiple formats. Learn more about Runs

​
Intel

Target-centric evidence workspace. The landing page lists targets with the age of their most recent probe, sweep, and import; each target has a detail page that surfaces the evidence end-to-end. Import findings from external tools (Garak, PyRIT, SARIF, scored-prompts, BIPIA), probe a model for IPI susceptibility, sweep document-context templates to pick the best one, and launch qai ipi generate with the recommended template directly from the target page. Learn more about Intel

​
Settings

Configure LLM providers, set defaults, and manage MCP infrastructure connections. All credentials are stored securely in the OS keyring. Learn more about Settings

​
WebSocket-Driven Updates

All active workflows stream updates to the browser via WebSocket. You’ll see real-time progress for:
  • Child run status changes (pending → in progress → completed/failed)
  • Finding discovery as modules complete
  • Result aggregation for multi-module workflows
The connection automatically reconnects if interrupted and resumes streaming from where it left off.
The Web UI and CLI share the same underlying database and configuration. Changes made in one are immediately visible in the other.

​
Browser Requirements

  • Modern browser with ES6 support
  • JavaScript enabled
  • WebSocket support for real-time updates
Tested on Chrome/Chromium 90+, Firefox 88+, Safari 14+.