A ZIP, read from the end
The archive keeps its index at the end, so 67 KB of a 40 MB file lists everything inside it.
Browser extension Chrome, Edge, Firefox
See inside a file before you download it
Right-click a file link, choose Open with Foretaste, and it opens. The bytes go from that website to your browser and are read there. Nothing is uploaded to us, because there is nowhere to upload it to.
0.17% of the file was read
read 67 KB of 40,013,838 bytes
All 10 files inside were listed from that. Opening one of them cost a further
97 bytes. Measured over HTTP against a server that logs every
request — counted at the reader, not estimated.
What it looks like
Two moments: the button appearing beside a file link on a page you were already reading, and what you get when you press it.
Downloads / 2026 / Course materials
course-materials.zip
ZIP archive
The page says 38.2 MB
Open Save
Hover or tap the marker and Foretaste says what the file is —
without fetching a single byte of it. Everything on that card was already
on the page.
course-materials.zip
ZIP archive 38.2 MB Files 842 Folders 73
67 KB read of 38.2 MB · 0.17%
Inside this archive
- lectures/ 412 files
- readings/ 287 files
- syllabus.pdf 1.2 MB
- timetable.xlsx 84 KB
- README.md 3 KB
Opening syllabus.pdf from inside costs
1.2 MB — that file, and nothing else in the archive.
The file tells you nothing until you open it
A link called attachment?id=18392. A 300 MB archive that might hold the slides
or might hold last year's. A spreadsheet from someone you half-know. You cannot tell from
the outside, so you download it, wait, find out, and now it is in your downloads folder
whether you wanted it or not.
Or you do not download it, and you never find out. On a work laptop or a shared machine that is often the sensible choice, and it is still the wrong outcome.
How you use it
-
Allow Foretaste on the site you are reading.
It arrives with access to no websites at all and asks for none. You grant one site at a time, from a button, and you can withdraw it just as easily.
-
Right-click the file link and choose "Open with Foretaste".
Or click the small marker Foretaste puts beside file links on pages you have allowed. Either way it is one action — no dialog asking what you meant.
-
Look at the file, then decide.
It opens in a panel over the page, or in a new tab if you prefer that. Save it to disk or close it and carry on reading.
What it opens
104 file extensions, grouped by what you get on the screen. The list below is generated from the extension's own detection table, so it cannot drift from what the code does.
Documents
.pdf .docx .odt .ott .pptx .odp .otp
Spreadsheets and tables
.xlsx .xls .ods .ots .csv .tsv
Archives
.zip .tar .gz .tgz .7z .rar .bz2 .tbz2 .tbz .xz .txz .lzma .zst .cab .iso .cpio .ar .lzh
Text, code and data
.txt .log .md .markdown .json .jsonl .ndjson .xml .yaml .yml .ini .conf .cfg .env .sql .js .ts .jsx .tsx .css .scss .html .htm .py .java .c .cpp .h .hpp .sh .bash .ps1 .rs .go .php .rb .sqlite .sqlite3 .db .parquet
Pictures and media
.jpg .jpeg .png .gif .webp .avif .bmp .ico .svg .heic .heif .mp3 .wav .m4a .ogg .flac .mp4 .webm .mov .m4v
Books, fonts and subtitles
.epub .ttf .otf .woff .woff2 .srt .vtt .ass .ssa
Calendars and contacts
.ics .ifb .vcf .vcard
What it will not do
Worth knowing before you install something rather than after. None of this is a roadmap item being coy — it is what the code does today.
- Word documents are a reading view, not a page layout. You get the text, headings, tables and images. You do not get pagination, headers or footers.
- A password-protected Word or Excel file cannot be opened at all. Those are encrypted differently from the file itself, and Foretaste carries nothing that can read them. A password-protected archive — ZIP, RAR or 7z — opens fine with a prompt.
- Spreadsheets show values only. No charts, no pivot tables, and formulas are not recalculated.
- TAR, GZIP, 7z and RAR are read whole. They keep no index to read instead, which is a
property of those formats rather than a shortcut. A TAR is at least not held whole:
it streams past, and one file inside a plain
.tarcomes back in a single request. - A database or a dataset opens from a few kilobytes. SQLite keeps its schema on its first pages and Parquet keeps its in a footer, so the tables, columns and types of a 500 MB file come out of one range read. Counting the rows exactly means walking every table, so it is a button rather than something that happens to you.
- Some sites will refuse. If a server blocks requests that did not come from its own pages, Foretaste says so and offers to save the file through the browser instead, which usually works.
- Most things are held in memory. A document, a spreadsheet or an image uses memory in proportion to its size, so Foretaste asks before opening anything over 50 MB. Listing an archive, paging a huge CSV and saving a file out of an archive do not: those stream, and a gigabyte comes out of a 4.8 GB archive without ever sitting in the tab. A million-row CSV turns a page at a time, and the last page costs what the first one did.
Your files never reach us
Every parser runs inside the extension, on your machine. Foretaste requests the file from the website it already lives on — the same request your browser would make if you clicked the link — and reads it there. No file, no file name and no page address is sent to a Foretaste server.
There is no analytics, no telemetry and no crash reporting, and a check in the build script
fails the release if any network call other than fetch, or any hard-coded
address outside a small allowlist, appears in the output.
Exactly what Foretaste can and cannot see, including every request it is capable of making.
Free opens it. Pro digs into it.
No format is behind the plan. Every file Foretaste can read, it reads for everybody — and saving to disk is free on every screen, including the ones where the preview failed.
Pro is depth on top: search every file inside an archive without unpacking it, filter a million-row table down to the rows you wanted, take the pictures out of a document, read rows out of a database, compare two files or two archives.
And it collects: every value of a field, across the whole file, as a list with how often each appeared. The file writes the menu. An XML document offers its own tags, a JSON file its own keys, a log its own fields, a CSV its own columns — so pulling every name out of an archive listing is a choice from a dropdown rather than a regular expression you had to work out first. What Pro costs.
Install it
The store listings are not published yet. Each page below tells you the state of that browser's build honestly, and what Foretaste asks your browser for.
Desktop browsers. Chrome extensions do not run on Chrome for Android — that is Google's limitation, not ours — and Firefox for Android is not claimed until it has been tested on one.