Chasei Workshop

Privacy policy

Chasei Workshop distinguishes 218 tools: 200 local, 13 local AI, and 5 network tools.

Local file and media processing

Selected files, names, previews, extracted text, results, PDF passwords, and game inputs remain in browser memory and are never sent to a server. Screen, camera, and microphone tools only use streams that you start and approve in the browser, and those streams stay local. Passwords are not stored, logged, or placed in filenames.

Document, font, and archive limits

Office and document conversion is limited to 64MiB per file and 500 PDF pages, and reconstructs text, tables, or page images, so complex fonts, formulas, animation, macros, and exact layout may change. Fonts are limited to 32MiB and 100,000 glyphs and convert TTF, OTF, WOFF, or WOFF2 locally. Archive conversion validates ZIP, TAR, TAR.GZ, TGZ, or single GZ input within 128MiB total, 64MiB per entry, and 5,000 entries; symbolic links, special entries, encrypted ZIP, 7Z, and RAR are unsupported.

PDF to HWPX scope

For input up to 64MiB and 100 selected pages, PDF text, numbers, symbols, ruled tables (including merged cells), two-column layouts, and figures become an HWPX document, and equations set with Hangul equation fonts are rebuilt as editable Hangul equations. Scanned pages can be read with optional Korean and English OCR in the browser or kept as images, and paragraphs and equations can be corrected side by side with the source before saving. Unusual fonts, handwriting, and complex layouts may be inaccurate. The selected PDF, extracted or edited content, and result are not sent to a remote conversion server.

Local storage

Favorite tool IDs, recent tool IDs, grid/list preference, theme, language settings, and radio and live-cam favorites, recent stations, volume, and preview preference are stored in localStorage. Versioned engines, models, and OCR language assets may be cached in Cache Storage or IndexedDB for reuse. Selected file names, contents, results, and game answers are not stored there.

World radio

Country and station searches run against a fixed list in the browser and are not sent to an external directory. Station selections are not reported to a click counter. For KBS, MBC, and SBS, our same-origin API fetches only a temporary stream address from the broadcaster’s official lookup service at play time. Audio is received directly from external stream servers, which receive IP and connection details. Streams may contain broadcaster advertising. Recording and redistribution are not provided.

World live cams

The curated list displays place, broadcaster, and original links without loading YouTube video until you press play; preview images are loaded from YouTube only if you turn previews on. YouTube then serves the stream directly and may receive IP and connection data or show broadcaster ads. This site does not count selections, record, or restream video.

Network requests

Radio, live cams and live TV connect to external broadcast servers. Official YouTube players for live cams and TV load only after playback is selected; YouTube may receive IP, browser and connection information. Opening an official viewing link connects to that broadcaster. Country/channel searches run locally; TV favourites and recents stay in this browser. The radio API only looks up official temporary stream addresses, and the TV API only looks up the current live video ID of a listed official YouTube channel; file-processing tools do not send their input to external servers. The SNS thumbnail downloader fetches YouTube thumbnails directly from i.ytimg.com in your browser; for TikTok, Threads, X and Instagram a same-origin API looks up only the public preview image address of the post link you enter, and the image then downloads from that site's image server straight to your browser. Those sites may receive IP and connection information; links and images are not stored. Separately, the Google AdSense code on visited pages may connect to Google servers. The site does not directly pass selected files, streams, names, extracted text, or result blobs to the ad code or tool API.

Analytics and ads

We do not use separate web analytics or session-recording scripts. The visitor count at the bottom of each page stores only a total per day on the server — no IP address, cookie or identifier. To count each browser once a day, the last counted date is kept only in this browser (localStorage). Google AdSense code is present; Google and advertising providers may receive your IP address, browser details, and visited page and may use advertising cookies. You can opt out of personalized ads in Google Ads Settings.

Google Ads Settings

Contact

Privacy questions: chasehwan2@gmail.com

Captions and transcription

Whisper Tiny and Base models download from public Hugging Face repositories on first use and may be cached by your browser. Download servers receive IP and connection information, not selected media or recognized text. Speech recognition runs in a local Worker. Teleprompter scripts and camera recordings remain in browser memory and are not uploaded or automatically saved.

Last reviewed: September 24, 2026