SlopShield

Scores YouTube videos 0-100 on a bullshit scale from their actual transcripts, right on the thumbnail.

Consuming anything online is a constant struggle to filter quality from clickbait - a quiet battle between you and the creators and algorithms trying to make few extra cents per click, your attention is all they need. Degenerative AI has aggressively tipped that battle: mass-producing low-quality slop that's indistinguishable from real content is now nearly effortless. This is an attempt to fight back. Instead of using AI to generate yet more content nobody asked for, it points the model the other way - letting it chew through the endless feed of garbage so you don't have to. Every video on your YouTube homepage gets a 0-100 bullshit score and a one-sentence summary of what's actually inside, judged from the transcript. A content-level firewall for the asymmetric battle against unsolicited slop. The score measures how badly a video breaks the promise its title makes: 0-20 LEGIT - does what the title says 21-40 MILD - delivers, with some rambling 41-60 MEH - real content buried in padding 61-80 BAIT - promises a lot more than it delivers 81-100 SLOP - filler stretched out for watch time HOW IT WORKS Your browser pulls the transcript from YouTube's own player API and sends it to the AI provider you pick (Anthropic, OpenAI, Google Gemini, or any OpenAI-compatible endpoint, including a local Ollama) with your own API key. No server, no tracking, nothing collected - the transcript goes from your browser to your provider and that's the entire data path. Scores are cached for 30 days; batched requests keep cost around $0.001-0.002 per video on the default models, and a local model is free. SETUP 1. Install, click the SlopShield icon. 2. Pick a provider, paste your API key, Save & test. 3. Reload youtube.com.