Web UI

Web UI — run Wild_Root_Prompt in your browser, no terminal needed

Wild_Root_Prompt ships a local browser interface for people who don't want to touch a command line. It runs a small Flask server bound to the loopback interface on port 7860 and streams tokens live via Server-Sent Events.

Nothing is hosted. The server is not reachable from other machines on your network, and it makes no outbound calls other than the optional web enrichment.

---

Launching it

How	Command / action
One click	Double-click WildRoot.command (macOS) or WildRoot.bat (Windows); run ./WildRoot (Linux)
Terminal	python3 prompt_expert_enhance.py web
Different port	python3 prompt_expert_enhance.py web --port 8080
Headless (no browser popup)	python3 prompt_expert_enhance.py web --no-browser

It opens http://localhost:7860 automatically. On macOS, if Chrome, Firefox, Brave, Edge, Arc or Opera is present, it opens that rather than Safari — Safari's HTTPS-Only mode hard-blocks plain http:// on loopback with no bypass, and a local server intentionally has no TLS certificate. See Troubleshooting.

---

The flow

1. Write your task in the input box — plain language, however rough.
2. Pre-process (optional) — a model rewrites your input into something structured and complete. You see the result and can edit it before continuing.
3. Pick a model — the dropdown lists everything installed locally.
4. Pick Quick or Full — one enhanced prompt, or the 12-section manifest.
5. Generate — output streams in live and auto-scrolls.
6. Copy or save the result.

Slash metacommands work here exactly as in the CLI: type /expert /tableau at the start of your task. See Slash Metacommands.

---

What the UI exposes vs. the CLI

The Web UI is deliberately the simpler surface. It covers the everyday path — pre-process, choose model, choose mode, generate, synthesize. The CLI additionally gives you technique ranges and bundles, draft mode, deep research, alternate backends, caching control, and scripting flags.

If you want the UI's convenience but the CLI's power, drive the server's endpoints directly — see REST API.

---

Persistence

Settings you change in the UI (models, mode, pre-processor, temperature) are saved to settings.json and shared with the CLI — the two interfaces are two front-ends over the same configuration. Details in Configuration.

---

If the page doesn't open

• Blank page or connection refused — the server may not have finished starting. Give it a few seconds and reload http://localhost:7860.
• Port already in use — start it on another port: python3 prompt_expert_enhance.py web --port 8080.
• "flask is not installed" — you installed with requirements-light.txt. Run pip install flask.
• Model dropdown is empty — Ollama isn't running, or no model is pulled. Start it, then ollama pull llama3.2:3b.

More in Troubleshooting.

---

Next: Examples · REST API — drive the same server from code · CLI Reference — what the UI doesn't expose.