MarkFlowy v0.100: Markdown Editor Rebuilt for Long Documents 2026
If you have a single Markdown file that holds a year's worth of archived X Articles, a Substack dump, or a Bluesky long-form export, you have probably hit the wall: VS Code takes 6 seconds to open it, Obsidian stutters when you scroll, and Typora quietly decides to re-render the whole document on every keystroke. The MarkFlowy v0.100.0 release on September 5, 2026, was a full editor-core rewrite aimed squarely at that workload. According to the release notes, a 2 MB Markdown file now opens in around 1 second on a recent laptop.
Why threadgrab is covering this: A typical archive exported from threadgrab (X Articles, threads, Bluesky long-form, LinkedIn newsletters) lands as a single .md file that easily crosses 1 MB when you bundle a month of a creator's posts. MarkFlowy v0.100 is the first desktop editor purpose-built for that file size without the typical Electron-based lag.
What Changed in MarkFlowy v0.100.0
MarkFlowy is an open-source AI-assisted Markdown editor (AGPL-3.0, 2,382 GitHub stars as of September 7, 2026) built on Tauri (TypeScript + Rust) and ProseMirror. The v0.100.0 release on September 5, 2026, was explicitly tagged as containing substantial breaking changes:
- Editor core rebuilt — the parser, document model, and WYSIWYG layer were rewritten to load large files faster.
- 2 MB in ~1 second — the headline metric from the release notes; previous versions were noticeably slower on files past ~500 KB.
- Refined WYSIWYG rendering — list nesting, task states, and inline formatting round-trip more cleanly through the Markdown source.
- Front Matter awareness — typing
---at the start of a document now creates a YAML front matter block instead of a horizontal rule (added in v0.100.1 on September 6). - Tab / Shift+Tab list indentation — nested content, task state, and undo/redo history survive the indent change.
The next-day v0.100.1 (September 6, 2026) patched three WYSIWYG-mode regressions that v0.100.0 introduced: Full Width toggle, the empty-paragraph placeholder hint, and the list-indentation behavior. If you upgrade, re-save existing documents once so the new internal representation takes hold.
Why the Long-Document Angle Matters for Social Archives
The threadgrab archive loop produces files that are exactly the workload MarkFlowy v0.100 was rebuilt for. A few examples from the kind of files real users open every week:
| Archive source | Typical file size | Section count |
|---|---|---|
| 10 creators × 1 month of X Articles | ~1.5 MB | ~400 posts |
| 5 creators × 1 year of Bluesky long-form | ~2.0 MB | ~600 posts |
| 50 threads concatenated into one .md | ~400 KB | ~50 threads |
| LinkedIn newsletter dump, one author, 6 months | ~800 KB | ~25 issues |
| Substack RSS flattened to Markdown | ~1.2 MB | ~30 posts |
All five of those fit the 2 MB MarkFlowy v0.100 benchmark. Editors that re-tokenize the whole document on open (VS Code with Markdown All in One, Obsidian with Live Preview, Typora's renderer) treat that size as a stress test. MarkFlowy's rewrite pushes more of the work to a smaller initial parse, which is the difference between waiting 6 seconds and waiting 1.
MarkFlowy vs VS Code, Obsidian, and Typora on 2 MB Files
Numbers below come from each project's own documentation and forum reports on the same hardware class (recent laptop, M-series Mac or Ryzen/Intel 12th-gen+). They are typical, not benchmark-perfect, but the ordering is consistent:
| Editor | Time to open 2 MB .md | Typing latency at 2 MB | Notes |
|---|---|---|---|
| MarkFlowy v0.100 | ~1 s | low | Tauri (Rust + TS), ProseMirror core |
| Typora | 2-4 s | medium | Live render is smooth after warm-up |
| Obsidian | 3-6 s | medium-high | Live Preview re-renders on each keystroke |
| VS Code (Markdown All in One) | 4-8 s | medium | Extensions add parse cost |
| Sublime Text | ~1 s | very low | No WYSIWYG; syntax-only highlighting |
Sublime Text is faster still because it skips rendering entirely. The trade-off is that you lose WYSIWYG preview, which is exactly what MarkFlowy's v0.100 rewrite tries to give back without paying the open-time cost.
How to Use MarkFlowy with a threadgrab Archive
The combined workflow is the same shape as the X Articles scraper article but with MarkFlowy replacing the editor step. It runs cleanly in under five minutes for a 1 MB archive:
Step 1 — Scrape the source
curl -s "https://threadgrab.com/api/profile/<handle>/articles?cursor=&limit=100" \
-o articles.json+ One call per author, paginated 20 at a time, returns named-entity body spans
− You still need to concatenate the JSON into a single Markdown file
Step 2 — Concatenate to one .md
jq -r '.articles[] | "# " + .title + "\n\n" + .body_markdown + "\n\n---\n"' \
articles.json > archive-2026-09.md
wc -c archive-2026-09.md # confirm size+ Standard Markdown with front matter, separators, and clean section breaks
− jq required; on Windows use python -m json.tool or a small Node script
Step 3 — Open in MarkFlowy v0.100
markflowy archive-2026-09.md # macOS / Linux binary
# or double-click the file from the File Tree panel+ Opens in ~1 second; scrolling stays smooth; the WYSIWYG mode updates without a re-open
− First-run on macOS requires xattr -cr /Applications/MarkFlowy.app until the app is signed
Step 4 — AI rewrite, section by section
# In MarkFlowy's AI panel, point at your DeepSeek or ChatGPT key,
# then ask: "rewrite section 12 to remove hedging, keep all facts."
# Review the diff in the same window, accept or revert per section.+ Section-scoped rewrites beat full-file paste-back-and-diff by a wide margin
− Requires you to bring your own API key; MarkFlowy does not bundle a model
Step 5 — Publish through md2rich or a static site
# Convert the polished Markdown to rich text for LinkedIn / Notion
curl -s "https://md2rich.com/api/convert?md_url=https://your-site/archive-2026-09.md"+ Markdown stays as the source of truth; rich text is a render
− LinkedIn still does not honor paste-with-formatting reliably; some manual fixing is normal
Who Should Switch to MarkFlowy v0.100
The rewrite is not for everyone. Three reader profiles get the most out of it:
- Social-archive maintainers. If you keep a single .md per author or per quarter, MarkFlowy is the first editor that opens those files instantly. The File Tree handles the per-author folder layout, and the in-editor AI panel lets you rewrite a section without leaving the document.
- Newsletter writers on long-form platforms. If your Substack or LinkedIn newsletter lives as a single Markdown source before it goes to the publishing platform, MarkFlowy's WYSIWYG mode gives you a live preview that does not stall at 2 MB. The Pandoc export added in v0.90.0 (DOCX / ODT / EPUB) covers the publishing-side hand-off.
- Writers who edit Markdown in VS Code today and feel the lag. If you have already disabled Markdown All in One or moved to Sublime because VS Code is slow on your archive, MarkFlowy gives you WYSIWYG without the VS Code weight. The keyboard-shortcut system is customizable, so the muscle memory carries over.
Three reader profiles should wait: anyone whose workflow depends on a specific VS Code extension, anyone whose team has standardized on Obsidian's plugin ecosystem (Dataview, Templater, etc.), and anyone who edits Markdown files under 200 KB (the speed difference is invisible at that size).
The Stack Behind the Speed
MarkFlowy is a Tauri app: the UI is TypeScript (3.86 MB of source) and the native side is Rust (506 KB of source). The editor itself is a ProseMirror build, with the v0.100 rewrite rebuilding the parser and document model to defer rendering until a section scrolls into view. The binary stays under 20 MB and ships a portable ZIP, an MSI, and an offline installer for Windows; DMGs for Apple silicon and Intel on macOS; and a Flatpak (via FlatPark) plus an install script that drops a .deb / .rpm / AppImage on Linux.
AI features are an in-app panel, not a separate service. MarkFlowy supports DeepSeek, ChatGPT, and Copilot today; you supply your own API key, the rewrite happens inside the editor, and the diff is shown in the same window. That is the practical difference for someone rewriting a long thread archive: section-by-section edits without leaving the document, instead of paste-back-and-eyeball.
FAQ
MarkFlowy is an open-source AI-assisted Markdown editor built on Tauri (TypeScript + Rust) and ProseMirror, available for Windows, macOS, and Linux. The v0.100.0 release on September 5, 2026, was a full editor-core rewrite with breaking changes; the headline result is that a 2 MB Markdown file now opens in about 1 second on a recent laptop. For X Articles archives, large wiki dumps, and long-form social posts joined into a single Markdown file, that is a meaningful shift because older desktop editors (VS Code, Obsidian, Typora) slow down or stall once a file crosses ~500 KB of nested structures.
A single author's X Articles archive can easily exceed 200 KB of clean Markdown when each post keeps its media list, named-entity formatting, and frontmatter. Stack a year of long-form posts from 10 creators and you are editing in the 2 MB range that MarkFlowy v0.100 now treats as a normal load. The same applies to Bluesky long-form posts pulled via the AT Protocol, LinkedIn newsletter exports, and Substack RSS dumps. Editors that re-tokenize the whole file on open become the bottleneck; MarkFlowy's rebuild pushes that work to a smaller initial parse, which is exactly the workload that scraped social archives produce.
On the same hardware, a 2 MB Markdown file with hundreds of sections typically takes VS Code 4 to 8 seconds to open with extensions like Markdown All in One enabled, Obsidian 3 to 6 seconds with Live Preview, and Typora around 2 to 4 seconds before it scrolls smoothly. MarkFlowy v0.100 reports around 1 second on the same file in their own testing on a current laptop. The headline gap is biggest on Windows, where Electron-based editors add the runtime overhead on top of the document parse.
Yes. threadgrab returns clean Markdown for X posts, threads, and Articles: each post keeps its bold, link, mention, and hashtag spans as standard Markdown, and a batch export can be concatenated into a single file. Open that file in MarkFlowy and the 1-second open speed holds for archives up to about 2 MB; beyond that, the editor still opens but typing latency starts to climb. The combined workflow is: scrape with threadgrab → open the .md in MarkFlowy → edit, lint, and rewrite → publish to Notion, Ghost, or md2rich for rich-text conversion.
v0.100.0 rebuilt the editor core, so some editing behaviors and interactions differ from earlier versions. The release notes flag the change explicitly and recommend re-saving documents that came from v0.90.x or older. v0.100.1, shipped the next day on September 6, fixes the Full Width toggle in WYSIWYG mode, restores the empty-paragraph Placeholder hint, and improves list indentation with Tab / Shift+Tab. New documents do not need any action; existing documents should be re-saved once after the upgrade to normalize the new internal representation.
Yes. MarkFlowy has a built-in AI panel that supports DeepSeek, ChatGPT, and other providers, and it can summarize, translate, or rewrite a long Markdown document in place. For the typical threadgrab workflow (scrape a long thread, ask an LLM to clean it up, republish the polished version), MarkFlowy keeps the rewrite interactive: you can ask the model to change one section at a time, review the diff in the same window, and only commit the parts you like. That is faster than passing the whole file to a chat UI and pasting the result back.
Try It on a Real Archive
The fastest way to feel the v0.100 speedup is to grab a small archive, concatenate it with jq, and open the resulting .md in MarkFlowy. A 200 KB file (about 50 X posts) will open in well under a second on any laptop made in the last three years. A 2 MB file (about 600 posts) is the workload where the rewrite pays off, and that is exactly the file size a creator's monthly archive produces when you keep the named-entity formatting.
Get the Markdown for your archive in one API call.
threadgrab returns clean Markdown for X posts, threads, Articles, and any public profile — no login, no install. Then open the .md in MarkFlowy v0.100 and edit at speed.
Open ThreadGrabRelated Reading
- X v11.90 Articles Scraper: threadgrab Long-Form 2026 — how threadgrab extracts the v11.90 article layout into Markdown that any editor can open.
- 2026 X Thread Archiving: 5 Tools Compared — the wider tooling landscape for saving X threads, where MarkFlowy slots in as the editor step.
- tweet.md Stable 2026: A Markdown Tool Retrospective — where tweet.md sits in the same archive pipeline, and how it differs from threadgrab's public-API approach.
- MRDown 2026: Verify AI Redrafts Before You Republish — the AI-meaning-drift check that pairs with MarkFlowy's in-editor rewrite panel.