Install the Tugboat feedback widget into my web app.

1. Add this script tag right before </body> (works with any framework — React, Vue, Angular, Svelte, Next.js or plain HTML; no npm install needed):
   <script src="https://tugboatapp.com/widget.js" data-project-key="pk_your_key"></script>
   Replace pk_your_key with my project key from the Tugboat dashboard (Projects → Settings).

Optional configuration:
- Keyboard shortcut: add data-hotkey="mod+shift+k" to the script tag (mod = Ctrl on Windows/Linux, ⌘ on Mac). Use data-hotkey="none" to disable.
- Manual trigger (no floating button): set the project to "manual" mode in Settings, then call Tugboat.open() from my own button/menu item.
- Audience targeting (for changelog & roadmap posts): identify the signed-in user with
  Tugboat.identify({ plan: "pro", role: "admin", segments: ["beta"] });
  or set window.tugboatUser = { plan, role, segments } before the script loads.
- Privacy: session replays already mask all inputs and auto-redact emails/phones/card numbers. Mark any sensitive region in my markup with data-tugboat-block (never recorded) or data-tugboat-mask (text masked).

Result: a floating "Feedback" button appears. Testers can send screenshots (with annotation), screen recordings with voice, voice notes and text — each with console/network logs, environment metadata and a session replay attached automatically.

Our docs are also machine-readable at /llms.txt and /llms-full.txt — point your AI agent straight at them.