DeWash - 祛惑认知防护盾

DeWash 是一款守护独立思考的认知防护盾，实时进行网页真实性核验、阻断恶意洗脑与春秋笔法话术、过滤不友善仇恨言论。

DeWash（祛惑）是一款专为独立思考者打造的隐私友好型浏览器防护盾。它不充当任何观点的裁判，而是作为您的认知辅助器，帮您剔除互联网碎片信息中的情感煽动与春秋笔法，回归客观事实。 【为什么选择 DeWash？】 现代网络环境充斥着大量预设立场的引导性叙事。很多文章并非在陈述事实，而是通过大量情感定性形容词（如“竟然”、“嚣张地”、“恬不知耻地”）来悄悄左右您的观点。DeWash 可以实时帮您把这些“带节奏”的话术找出来，并给出 AI 诊断分析。 【核心功能防线】 1. ⚖️ 主观感情色彩与“春秋笔法”警示：自动放行使用的是中立语调的客观事实陈述，专门标记并警示带有主观评判、立场偏颇的修饰词和定性用语。 2. 🔍 事实真伪核验：自动匹配公认科学常识与权威报道，拦截和标识反常识谣言或伪科学断言。 3. 🛡️ 仇恨言论与地域歧视中和：高亮并过滤恶意人身攻击、极端地域对立与性别对立言论。 4. ⚙️ 三大触发模式：支持“按需手动”、“自动白名单”及“全局模式”自由切换，完美控制 AI 流量与 Token 消耗。 5. 🎛️ 可信度预警阀门：配备了可调的预警阈值滑块（Trust Score Slider），低于该阈值的分数才会进行网页划线，彻底免除过度警告噪音。 6. 🌐 搜索引擎一键跳转追溯：AI 将语义化提取存疑内容的 1~2 个核心实体，支持一键跳转搜索引擎进行交叉对比验证。 【零隐私风险承诺】 DeWash 采用 100% 客户端本地运行模式。您需要自行准备大模型 API KEY（如 OpenAI, Google Gemini, DeepSeek 等）。所有的秘钥、网页扫描文本均安全加密存储于您本机的浏览器环境（chrome.storage）中，不流经任何第三方中转服务器，绝对保障您的商业机密与阅读数据隐私。 --------------------------------------------------------------------------------------------------------------------------------------------------- DeWash is a privacy-first browser extension designed to protect your cognitive sovereignty. It serves as your cognitive assistant, helping you strip emotional framing and subjective spin from online texts to reveal the core objective facts. 【Why DeWash?】 Modern media is filled with biased narratives. Instead of presenting objective facts, articles often use emotionally loaded qualifiers (e.g. "shockingly", "shamelessly", "viciously") to manipulate reader perspectives. DeWash highlights these subjective framing devices in real-time and provides detailed AI cognitive diagnosis. 【Key Defenses & Features】 1. ⚖️ Bias & Framing Warning: Automatically permits neutral factual statements to reduce noise, flagging only loaded words, moral judgments, and biased framing. 2. 🔍 Fact Verification: Checks against scientific consensus and authoritative reports to instantly flag fake news, pseudoscience, and rumors. 3. 🛡️ Hate Speech Neutralizer: Intercepts and flags toxic personal attacks, group stereotypes, and xeno/geographic hostility. 4. ⚙️ 3 Flexible Trigger Modes: Switch between "On-Demand Manual", "Auto Whitelist", and "Global Scan" to avoid accidental token consumption. 5. 🎛️ Trust Score Threshold Slider: Fine-tune checking sensitivity directly from both the popup and options tab. Flags text only when its trust score falls below your setting. 6. 🌐 Search Verification Lookup: AI extracts semantic entities from flagged claims for one-click search validation on Google, Bing, or Baidu. 【100% Privacy & Local Storage】 DeWash is a pure client-side shield. You must provide your own AI API KEY (OpenAI, Google Gemini, DeepSeek, or any compatible endpoint). All keys and scanned text are encrypted and stored strictly in your local browser (chrome.storage). No third-party servers are involved, ensuring absolute data privacy.