ZIP to Markdown: One Upload, the Whole Archive Converted
Converting documents one at a time is fine until there are twenty of them. Then it's twenty uploads, twenty waits, twenty copy-pastes, and by file fourteen you've lost track of which ones you've already done. That's the problem this page exists to remove.
Zip the folder. Upload the one archive. Every supported document inside gets converted in a single pass and comes back as Markdown, with each file's own structure intact and clearly separated from the next. It's a batch job disguised as a file upload — and it's the same free tool as the rest of the suite. No sign-up, 50 MB limit, nothing kept afterwards.
What Actually Happens to Your Archive
The archive is unpacked, the file tree inside is walked top to bottom, and every file with a supported extension goes through the same conversion it would have got on its own dedicated page. A PDF inside your ZIP is treated exactly like a PDF uploaded directly. A spreadsheet becomes tables. A deck becomes slide-by-slide sections. Then the results are stitched together into one Markdown document.
Two details matter more than people expect. First, the folder structure inside the archive is
preserved as context — you can tell that 2024/q3/board-pack.pdf came from a different
place than archive/old/board-pack.pdf, which matters enormously when filenames repeat across
directories, and they always do. Second, the archive is unpacked in place, not flattened into a
pile. The hierarchy someone built when they organised those files is itself information, and
throwing it away makes the output harder to reason about.
Mixed-format archives are the normal case, not the awkward one. A ZIP containing four PDFs, two Word documents, a spreadsheet, and a PowerPoint deck converts in one go, each file handled by the logic appropriate to its own format.
What Converts and What Gets Skipped
Anything the converter handles as a standalone upload also works inside an archive — PDF, DOCX, PPTX, XLSX and XLS, HTML, CSV, JSON, XML, EPUB, RTF, MSG, and images. If you want to know exactly how a given type behaves, the individual pages go into detail: PDF to Markdown, Word to Markdown, PowerPoint to Markdown, and Excel to Markdown cover the four most common things people find in a work archive.
What gets passed over is everything with no text to give. Videos, audio, fonts, compiled binaries, installers, and proprietary formats outside the supported list contribute nothing and are left alone. Images are the case worth flagging, and it catches people out. Uploaded on their own, image files get read properly — text in a screenshot or a photographed page comes back as text. Inside an archive they don't. The batch pass records that the file was there and moves on, so a ZIP of screenshots converts successfully and still hands you filenames instead of the words on the screens. If those words are the point, pull the pictures out and run them through image to Markdown directly.
Nested archives — a ZIP inside your ZIP — are the other thing to check. If your export arrived as an archive of archives, unpack one level yourself and re-zip the contents before uploading. Two minutes of housekeeping, much better output.
Why Markdown Is the Right Call for a Batch
With a single document, choosing Markdown over flat text is a judgement call. With twenty documents in one output, it stops being close.
The reason is boundaries. When several files are concatenated into one body of text, the hardest thing for both you and a language model is knowing where one document ends and the next begins. Markdown headings give you that for free — each file announces itself, and everything under it visibly belongs to it. Ask a model "which of these contracts has the shortest notice period" and it can answer per-document rather than blurring clauses from three agreements into one confident, wrong answer.
Per-file structure survives too. The spreadsheet's tables stay tables. The deck's slides stay separate sections. The report's clause numbering stays intact. That's the whole argument for batch converting documents to Markdown rather than to raw text: structure is what makes a multi-document context navigable instead of just large.
The exception is bulk analysis — embeddings, search indexing, keyword work across a corpus, anything where you want one flat pile of prose and syntax characters are noise. That's ZIP to text, and the toggle at the top of this page switches to it while keeping the archive you've already selected.
The Archives People Actually Upload
- Email attachment bundles. Someone sends "everything for the review" as a ZIP of nine attachments. Convert the lot, read the summary, decide which two you actually need to open properly.
- Data exports. Google Takeout, Slack workspace exports, Notion workspace dumps, and support-desk exports all arrive as archives full of nested folders. This turns a directory tree you'd otherwise click through for an hour into something readable in one pass.
- Course material packs. A module's worth of lecture slides, handouts, and reading lists in one download. Convert it, then ask for a study guide built from the actual materials rather than the model's general knowledge of the subject.
- Client handovers and due diligence packs. A data room export, a vendor's document bundle, a project handover folder — the whole thing becomes searchable text in one step, with folder paths telling you the provenance of each piece.
- Grant and tender submissions. Multi-document responses with appendices, where you need to check consistency across files that were written by different people.
- Personal archives. Old backup folders where you can't remember what's in them. Convert first, dig second.
The 50 MB Line, and Living Under It
The limit applies to the archive you upload, not to individual files inside it. That's usually generous, because ZIP compresses text-heavy formats well — a folder of Word documents and spreadsheets can shrink dramatically, so archives with a lot of files still often fit comfortably.
What blows past the limit is media. Video, audio, and high-resolution images barely compress and will consume the entire budget while contributing nothing to the output. If your archive is over the line, strip those out and re-zip — you lose nothing you were going to use. Split genuinely large document sets into two or three archives by folder and convert them in sequence.
The other budget worth watching is the token counter under the output. Fifty documents converts fine and still won't fit in a small model's context window. Knowing the number before you paste is better than finding out from an error, and it tells you whether to send the whole batch or work through it in chunks.
If Your ZIP Is Really a Code Project
Worth saying plainly, because it comes up constantly. If you've zipped a repository or a source folder, this
is the wrong tool. Use the local directory
converter instead — it reads a folder straight off your machine, shows you a checkbox tree of the file
structure, and lets you pick exactly which files go into the output. That means you can drop
node_modules, build artefacts, and lock files before they eat your entire context window,
which is not something a blind archive conversion can do for you.
Project already hosted somewhere? Skip the zipping entirely — the GitHub repository to text converter pulls the tree directly from the URL.
Frequently Asked Questions
How do I convert a ZIP archive to Markdown?
Upload it above and each supported file inside is converted to Markdown and concatenated, with the archive's folder paths kept as headings. Free, 50 MB per archive, no account. The result is one .md document where you can still see which section came from which file.
Do the file paths become headings?
Yes, and that is the reason to choose Markdown over flat text for an archive. Each file's content sits under a heading naming its path inside the ZIP, so the output stays navigable and a model asked about a specific document can locate it instead of guessing from context.
Are nested archives handled?
A ZIP inside a ZIP is treated as another supported file and unpacked in turn. Useful for exports that bundle per-month archives inside a yearly one. Very deep nesting is worth avoiding — the output gets long and the heading levels bottom out at six, after which structure stops being expressed.
What order do the files come out in?
Directory traversal order, which means files group by folder rather than by relevance. For an archive with a meaningful layout that is exactly right. For a flat dump of two hundred files with generated names, the order is arbitrary and you will want to search the output rather than read it top to bottom.
Is there a limit on the number of files inside?
The practical constraint is the 50 MB archive cap rather than a file count, but keep an eye on the token counter under the output. A few hundred documents can exceed most model context windows several times over, so it is often better to convert the subset you need than to produce a corpus you cannot paste anywhere.
The Rest of the Toolkit
File2Txt handles all thirteen supported formats from one place if you'd rather not pick a page. And the guide to preparing files for LLMs makes the broader case for converting before you prompt.
Repo2Txt is built and maintained by v12hero, an independent developer building privacy-first native and web apps.