Free · No installer · Never touches the network

You don't need to remember the filenameOne word from inside the document is enough

That penalty clause in some contract. That one figure in a report. CHATDA indexes the text inside HWP, PDF, Word and Excel files, so you find them without opening a single one.

Download for Windows See what's different

1 million files indexed in 1.1 s · sub-millisecond search · 1.9 MB executable · no installer, no account

CHATDA window. Searching for a Korean word returns six files including a legacy HWP document, each row showing the matching sentence highlighted in yellow.

CHATDA window in dark mode, showing search results with matching passages highlighted.

A real screenshot. The yellow parts are passages found inside the documents.

The line the market leader wrote about itself

Finding a file by name is a solved problem. The trouble starts right after that.

“File content is not indexed, searching content is slow.”
— Everything official FAQ, voidtools.com/faq

The files people genuinely cannot find are the ones whose names they no longer remember. “Which contract had that clause in it?” — a filename search will never answer that. That is where CHATDA starts.

Three things, done properly

Rather than adding features, we opened the three places where office search actually gets stuck.

Content

It reads inside HWP too

Hangul Word Processor 5.0 and 3.0, HWPX, PDF, Word, Excel, PowerPoint, ODF — and old .txt files saved in CP949.

The matching passage is cut out and shown right in the result row, so you never open a file to check.

One query reaches a contract, a meeting note and page 15 of a report at once

Languages

Search that knows the language

Korean initial consonants, keyboard-layout mistakes, one-character typos, and macOS decomposed filenames all resolve to the same name.

The same principle covers 12 languages — Chinese pinyin initials, Japanese romaji, and Cyrillic, Arabic and Devanagari transliteration.

Type bgs for 报告书 · houkoku for ほうこく · godovoy for Годовой

Private

Nothing leaves your PC

No network calls. No account, no sign-in, no cloud upload. The interface binds to 127.0.0.1 only.

It is read-only by default; moving or renaming files has to be switched on deliberately.

Zero third-party libraries · one executable · delete it and it's gone

It stays light

A heavy indexer is one you eventually turn off. Size and speed were design constraints from the start.

1.1s

to index 1,000,000 filenames

<1ms

per search (average)

99MB

resident at 1,000,000 files

1.9MB

one executable, no installer

Measured on a synthetic 1,000,000-file corpus (12 languages, evenly distributed) on 2 cores. Real numbers vary with your disk and file mix.

Duplicatesfound by content, not by name
Query syntaxext:hwp size:>10mb date:7d, regular expressions
AI accessa built-in MCP server lets tools like Claude find and read your documents
12 languagesinterface and search alike (ko en ja zh ru ar hi de fr es pt id)
It learnsfiles you open often rise; the record never leaves your machine
Command lineCSV and JSON output to pipe into PowerShell

Try it once

Download, double-click, done. If you don't like it, delete one file.

Download for Windows