Plot Armor

Chrome extension that blurs likely spoilers while you browse. It combines local story-graph matching with an OpenAI semantic check so spoilers can still be caught when a show title is not spelled out.

Tech stack

• Chrome Extension (Manifest V3)
• JavaScript
• OpenAI Chat Completions API
• TMDB API
• Chrome Storage API (sync + local)
• MutationObserver + IntersectionObserver

Demo

_{Search TMDB, pick a suggestion, then Add.}

_{Matched content is blurred; click to reveal.}

Setup

1. Clone this repo.
2. Open chrome://extensions, enable Developer mode, Load unpacked, select this folder.
3. Configure API keys — right-click the extension → Options (or use Extension options on chrome://extensions):

• OpenAI API key — semantic spoiler detection
• TMDB Read Access Token — search and metadata

Legacy: copy .env.example → .env (gitignored) for OpenAI, TMDB, and false-positive ingest. Or set API keys in Extension options.

Usage

1. Open the popup, search TMDB, select a row (Add stays disabled until you do), click Add.
2. Toggle shields per title, or use pause all / enable all.
3. Browse as usual — Reddit, Wikipedia, X, etc. Blurred blocks can be revealed per click. Use not a spoiler? on a revealed block, or Options → false positives, to report mistakes.

Docs

• Developer guide — architecture, detection pipeline, debugging, fixture tests

Known limits

• Adds must come from TMDB suggestions (no free-text or manual keyword lists yet).
• Default LLM escalation favors recall over cost — more API use than a local-only gate.
• Detection and context quality still vary by site layout and model behavior.
• Protected-list updates can briefly re-scan and flash existing blurs on some pages.

Safety

Page text snippets are sent to OpenAI for analysis. Do not commit API keys or share screenshots that expose credentials.