Flags loaded, sensationalized wording in articles and swaps it for neutral phrasing, powered by Gemini. Redline reads the Persian-language news article you're currently viewing and highlights wording written to sway rather than inform — then suggests a neutral replacement, with the reason visible on hover. It doesn't touch the facts, only the word choice. TWO PASSES, ONE JOB EACH Phase 1 is instant and fully offline: a curated, symmetric dictionary of ~200 unambiguous label swaps (things like "rioter" → "protester") applies the moment the page loads. No network request, no API key needed. Phase 2 is optional and needs your own free Gemini API key: it reviews full sentences and paragraphs for the kind of bias a fixed word list can't safely judge — loaded adjectives, unverified claims stated as fact, a sentence built on a slanted premise. Results appear both inline and in a compact report panel. BUILT TO BE CHECKED, NOT TRUSTED BLINDLY The instant-pass dictionary ships as a plain, readable file — every entry and every reason is inspectable, and it's deliberately symmetric: labels drawn from state-aligned coverage and from opposition/diaspora coverage both appear in it. A handful of swaps that might look tempting are deliberately left out because they'd assert a contested position rather than remove one (for example, treating "Zionist" as synonymous with "Israeli"). The reasoning is documented in the source. INSTALL
Comments