Centralized MTurk HIT Catcher

I’m building a single spot where all my Amazon Mechanical Turk HIT catching happens. The vision: a lightweight PHP back-end that talks to a Tampermonkey/Greasemonkey userscript running in the browser, so I no longer juggle multiple browser tabs or scattered scripts.

Core goal
The system must auto-catch HITs for me. It should do this intelligently—dynamic interval changes, pausing or speeding up when the queue is dry or suddenly busy, and prioritising based on requester, reward, or any keyword rules I define on the fly.

How I picture the flow
• A small web dashboard (PHP + MySQL or SQLite) lets me add keywords, set per-requester priority, view logs, and toggle the catcher.
• The userscript polls the PHP API, fetches my current rules, and fires the accept call the instant a matching HIT shows up.
• Each accept attempt, success, or block gets logged so I can review performance later without digging through console output.

What I need from you
1. Clean, commented PHP back-end (plain PHP is fine; feel free to pick a micro-framework if it speeds you up).
2. Tampermonkey/Greasemonkey userscript that plugs straight into MTurk, handles authentication cookies, and respects MTurk rate limits.
3. A simple SQL file to create any tables.
4. A concise README so I can drop this onto a shared host, import the database, install the script, and start catching.

Acceptance criteria
• The catcher must successfully accept at least one live HIT matching a test keyword while dynamically adjusting its polling interval.
• Turning rules on/off in the dashboard instantly affects the userscript without a page refresh.
• All activity (hits found, accepted, missed, errors) is recorded and viewable in the dashboard.

If anything in the outline could be improved or streamlined, let’s discuss it before you dive into code.