尚未检测到更新。GitHub 发布版本后会显示在这里。
按最近已完成的 8 小时观测回合间的 GitHub 星标增量排序
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权，自行管理介绍，无需管理员批准。
登录并生成验证文件，在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件，即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录，将前往登录页面。登录后请回到此处，再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify，即可验证所有权并自行管理介绍。
一分钟完成验证Remove invisible-character watermarks from AI-generated text. Runs entirely in your browser.
翻译准备中
取自仓库 README
正在加载完整 README。 以下为已保存的摘录。
Paste AI-generated text, see the invisible characters hidden in it, and copy a clean version.
The cleaner runs entirely in the browser, so pasted text never leaves the page. The site is a
static export and uses Google Analytics to measure page visits.查看完整 README ↗Paste AI-generated text, see the invisible characters hidden in it, and copy a clean version. The cleaner runs entirely in the browser, so pasted text never leaves the page. The site is a static export and uses Google Analytics to measure page visits.
It removes invisible-character watermarks: zero-width spaces and joiners, BOMs, soft hyphens, bidi overrides, variation selectors, Unicode tag characters (which can spell out hidden messages), private-use/noncharacter/reserved code points and exotic spaces. It keeps the invisible characters real text needs (emoji ZWJ sequences, Persian/Indic joiners, flag tag sequences, RTL marks).
It cannot remove statistical watermarks such as SynthID-Text, which live in word choice rather than in any character.
skills/remove-ai-watermarks is an Agent Skill that gives coding
agents the same cleaner. Install it with the skills CLI, which supports
Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot and more:
npx skills add heyyarun/i-hate-watermarks
Or copy the folder by hand into ~/.claude/skills/ (Claude Code) or ~/.agents/skills/
(Codex). Then ask your agent something like "check essay.md for hidden AI watermark
characters and clean it". The skill runs scripts/watermarks.mjs, a single file with no
dependencies that needs Node.js 18 or newer:
node skills/remove-ai-watermarks/scripts/watermarks.mjs inspect essay.md
node skills/remove-ai-watermarks/scripts/watermarks.mjs clean -i essay.md
pbpaste | node skills/remove-ai-watermarks/scripts/watermarks.mjs clean | pbcopy
watermarks.mjs is generated from packages/skill and committed so the skill folder works
on its own. After changing packages/core or packages/skill, rebuild it with
corepack pnpm build:skill; a test fails if the committed file is out of date.
packages/core: the cleaning engine (TypeScript, no dependencies). Port of
watermarks-remover's Layer A engine.packages/skill: the skill's command-line front end, bundled into
skills/remove-ai-watermarks/scripts/watermarks.mjs.skills/remove-ai-watermarks: the installable skill (SKILL.md plus the bundled script).apps/web: the Next.js site (output: "export").Uses pnpm through corepack (bundled with Node), so pnpm does not need to be on your PATH.
If it is, plain pnpm <script> works too.
corepack pnpm install
corepack pnpm dev # http://localhost:3000
corepack pnpm test # engine and skill tests, incl. parity with the Python original
corepack pnpm typecheck # works from a clean checkout; generates Next route types first
corepack pnpm build # static site in apps/web/out
corepack pnpm build:skill # skills/remove-ai-watermarks/scripts/watermarks.mjs
Both scripts need uv. Fixtures also need a checkout of the reference repo.
cd packages/core
corepack pnpm gen:names # src/unicode-names.generated.ts
corepack pnpm gen:fixtures /path/to/watermarks-remover # test/fixtures/parity.json
MIT. See LICENSE and NOTICE.