This PR is a bounty claim implementation of #2169.

Features of the CLI/how it works:

• First it asks for the url of the website to screenshot. The URL is validated before passing it to Firecrawl.
• The screenshot is extracted from Firecrawl API's scrape endpoint
• The user will select a preset or a custom editing prompt
• Prompt will be passed to Gemini 2.5 Flash (aka Nano Banana) for editing
• The user will now have these options:

1. Further edit the edited image with custom prompts (apply additional edits using the most recent version as reference)
2. Redo the edited image with custom prompt (try a different prompt using the same reference image that created the current version)
3. Redo the whole thing with a different preset (apply any preset or custom prompt to the original screenshot)
4. Start over with new url
5. Exit

It continues as a loop until the user manually exits. It allows for step by step editing, and each generated image is saved in the cwd.

Summary by cubic

Adds a Python CLI example that screenshots a website with Firecrawl and edits it using Gemini 2.5 Flash. Supports presets or custom prompts, iterative edits, and versioned PNG output.

• New Features

  • Interactive CLI: URL validation/normalization, screenshot capture via Firecrawl.
  • 12 style presets + custom prompts, with clear editing guidance to the model.
  • Edit loop: continue editing, redo with same reference, redo from original, start over.
  • Auto-save to PNG in the current directory with timestamps and versioning.
  • Included README, .env.example, and requirements for quick setup.

• Dependencies

  • Example-only: google-genai, firecrawl-py, Pillow, questionary, python-dotenv, validators, requests (and related pins).