> Platform Note: Ferrite has been primarily developed and tested on Windows. While it should work on Linux and macOS, these platforms have not been extensively tested.
Neat! Lately on Windows I've felt like a 2nd class citizen.
> AI Disclosure: This project is 100% AI-generated code.
This is cool. I was hoping to see progress coming from Zed (e.g. because Tree-sitter → https://github.com/tree-sitter-grammars/tree-sitter-markdown) but it's exciting to see this.
I'm a heavy Obsidian user, and I love it, but I'd love to see real alternatives focused on foundations.
It would be interesting to know more about the end-goal if any.
I am aware of the current issues with open source licensing, but for my needs I don’t trust the elastic style licensing, especially when it claims to be open source but I can’t fork it to protect myself from a future rug pull situation.
I currently use Dendron in VS Code. Dendron is basically abandonware at this point because it couldn’t be monetized, but because it’s Apache licensed, I can fork it if I want, and continue to use it until something better comes along, or even modify it for my own needs.
It’s very hard to be successful financially in this space. Notion did it at the right time, but they are targeting enterprises who are willing to give their data to them, not individuals who want to run their own setup.
Maybe you can compete with Notion, but I’m not willing to put my stuff in a system that may not be around in a couple years, and I don’t have a license for.
You need something "more" on the website before you ask people to create an account. "Team workspace that stays fast" isn’t clear enough for me, at least. What is a workspace? What does the interface look like? Is it in the browser? Is it an app?
People will go "what is this?", "huh, I’m not gonna make a user for this, can’t tell what it is". Those are my 2 cents.
Disclaimer: I'm not your target audience, I don't care about collaboration or performance.
- There's a heavy emphasis on performance. Are you sure customers care about that more than real time collaboration and self hosting? (I don't think they care about CRDTs.)
- If I am experiencing pain because eg my Notion wiki is too big and is having serious performance issues, what I want to hear immediately is how you are going to help me migrate from Notion to your solution. Notion has a feature to export an entire workspace; can you ingest that and get me spun up with your product?
- If I hear something is open source I expect to be able to try it out immediately without logging in. It looks like you can do that but when you hit "Get Started" it puts you into a registration flow.
- You might take a look at how Zed is marketing themselves, they have a similar pitch (performance + realtime collaboration). The first thing they try to show you is a video where they demo the product and show how fast it is. (I think they focus too much on performance though.)
- The frontend is a web app right? If possible rather than a video, embed the interface in your landing page. If possible, let them share their document and try out collaborating on it with someone or with another browser tab. Give them an opportunity to be impressed.
I respect anyone who posts their work. Best of luck.
> There's a heavy emphasis on performance. Are you sure customers care about that more than real time collaboration and self hosting?
- Good point, I'll find out
> Notion has a feature to export an entire workspace; can you ingest that and get me spun up with your product?
Yes, I'm almost done with this feature
> If I hear something is open source I expect to be able to try it out immediately without logging in. It looks like you can do that but when you hit "Get Started" it puts you into a registration flow.
I link to that elsewhere in the page: https://hyperclast.com/dev/ I'll look into making this more prominent.
> You might take a look at how Zed is marketing themselves
I use Obsidian a lot, but very few extra features or plugins. My first impression is that I don’t get what you’re making from the website. Any tool worth using in this space (which I vaguely understand to be using large collections of Markdown and/or realtime multiediting) is fast. Obsidian is fast. Zed is fast. It’s table stakes for the kind of person who would use this already.
Is it just Zed + Obsidian? A good knowledge base that scales well and uses plain markdown, but has the fancy multi edit stuff?
Thanks! The end-goal is a fast, native Markdown editor that "just works" - no Electron, no web tech, instant startup. v0.3.0 will extract Mermaid as a standalone crate and build a custom text editor widget to unlock features egui's TextEdit blocks (proper multi-cursor, code folding). Long-term: potentially extract the editor as a headless Rust library since that's missing in the ecosystem. See ROADMAP.md for details
My impression was that everyone uses their $EDITOR and integrates languages support via plugins. The only exception to this rule I know is Emacs (org mode). I really doubt a standalone md editor will get traction, no matter how good it is.
Market exists: Obsidian has 1M+ users, Typora is popular, iA Writer has a loyal following. These aren't VS Code users who wandered off — they're writers, PKM enthusiasts, and note-takers who find IDE-style editors overwhelming for prose.
Different audience: Developers might prefer VS Code + Markdown Preview Enhanced. But Ferrite targets people who want a focused writing tool, not a general-purpose editor that happens to support Markdown. Think "writing app" vs "code editor with Markdown support."
Native advantage: Most Markdown tools are Electron (Obsidian, Typora, Mark Text). Ferrite offers instant startup, lower RAM, and native performance — appeals to the "I want my tools to feel fast" crowd.
You might be right that it won't achieve mass adoption. But there's a niche for "Obsidian but native and lighter" that I think is underserved.
My impression was that Obsidian is more than an editor: personal wiki, todo tracker, database, etc..
The currently offered feature list in Ferrite — code blocks, mermaid — suggests you are targeting developers or tech people here, hence, not really iA Writer... Typora — never heard of it, can't comment.
Anyway, thanks for seeing this as skepticism, and not criticism.
With my comment, I tried to subtly suggest that there should be more to it, than an editor.
Completely non-accusatory, just wondering. Did you write this post using an LLM? I sort of feel the typical "voice" if LLM writing here and wondering if I should calibrate myself a bit in this.
Open source purity is problematic. The OSI was established by the hyperscalers, who are decidedly not open source either.
Purely "OSI-approved open source" mandates having no non-commercial or non-compete clause, which means anyone can come in and bleed off profits and energy from the core contributors of open source projects. It prevents most forms of healthy companies from existing on top.
We shouldn't be allergic to making money with the software we write - life is finite and it's more sustainable over the long term to maintain software as a job.
The new "ethical source" / "fair source" licenses that have been popping up recently [1, 2] give customers 100% use of the code, but prevent competitors from coming in and stealing away the profits from running managed offerings, etc. (I wish Obsidian were this, but it's fully closed. Still, I do not admire them any less for this choice. We venerate plenty of closed creators - it's silly to hold software to a different standard.)
AWS profits hundreds of millions a quarter off of open source developed by companies thinking they were doing the right thing. AWS turned these into a proprietary managed solutions and gave nothing back to the authors. The original wind up withering and dying. AWS isn't giving back, they're just hoovering up.
Obsidian being closed means the core authors are hyper focused and can be compensated (even if it's not much). It's not like they can rug pull us - the files are text files, we can use old versions, and if they did piss us off I'm sure someone would write an open source version.
Fully agree that pushing OSI is just posturing. After all, Amazon/Google/Facebook have made literal billions by commercializing open source software. I release stuff on MIT all the time (for things I'm okay with people poaching) but I'd argue the only "pure" OSS license is GPL, which comes with its own problems (and, as we all know, it infects everything it touches).
The problem with FSL is that it hasn't been tested in the courts yet (afaik), so it's a bit of a gamble to think it'll just "work" if some asshole does try to clone your repo and sell your work. Maybe it's a decent gamble for a funded startup with in-house counsel, but if you're just one guy, imo keep stuff you want to sell closed-source, it's not that big of a deal. We've been doing just that since the 70s.
Shamelessly plugging my app Octarine (https://octarine.app) for users that may want a more WYSIWYG editing experience while storing all notes on device in markdown, which is also written in Rust (Tauri), and NOT vibe coded :)
Will need a magnifying glass to see the text on the screenshots.
I find it makes sense to take screenshots in a window big enough to show what's going on, but no bigger. This means probably not full screen, or maximised, especially if you're running at a very high resolution. If there's a lot of dead/empty space in the window that's a signal it's too big. This way you guarantee the screenshots are readable without zooming in, on smaller displays than your own, for example mobile.
egui is fantastic for rapid prototyping - immediate mode makes state management simple. Main limitation: TextEdit isn't designed for code editors (no multi-cursor, can't hide folded text). v0.3.0 will replace it with a custom widget. The default styling does scream "egui" - spent time on custom theming to avoid that
One major downside of native rendering is the lack of layout consistency if you’re editing natively and then sharing anywhere else where the diagram will be rendered by mermaid.js.
This is a perfect use case! The v0.3.0 crate will have:
- parse() → AST
- layout() → positioned elements
- render_svg() → SVG string
- render_png() → via resvg (no browser needed)
CLI usage would be something like:
mermaid-rs diagram.mmd -o diagram.png> # or pipe from stdin> cat diagram.mmd | mermaid-rs --format svg > output.svg>
For your mark integration, you'd be able to call it as a subprocess or use it as a Rust library directly if you're building in Rust.
If you want to follow progress or have input on the API, feel free to open an issue on the repo!
Valid point! Native rendering won't be pixel-perfect with mermaid.js. The trade-off is speed and no JS runtime. For documents staying in Ferrite, it's great. For sharing, we're adding SVG export in v0.3.0 so you can use mermaid.js for final renders if needed.
I happily paid money for Typora, which does roughly the same thing for just Markdown without support for JSON, Yaml (that I know of). This feels like a ripe space, especially with LLMs eagerly outputting reams of parseable text with embedded diagrams.
The price of a fancy burger doesn't seem all that unreasonable for a piece of software one finds even moderately useful (of course, depending on your local exchange rate that may be more or less true)
Thanks! Typora is great - Ferrite aims for similar polish but with native Mermaid, structured data support (JSON/YAML/TOML tree viewer), and the pipeline feature for shell integration. And it's open source!
Ha! I appreciate the compliment (I think?). To be transparent: yes, AI tools were used during development — they're fantastic for boilerplate, documentation, and exploring unfamiliar APIs.
But this wasn't "2 sessions" — Ferrite has been in development for months with ~30,000 lines of Rust across 50+ modules. The Mermaid renderer alone is ~6000 lines of layout algorithms (Sugiyama-style graph layout, sequence diagram activation tracking, nested state machines, etc.).
AI helped ALOT, but there's no "generate full app" prompt that produces working text editors with native diagram rendering, rope-based text buffers, and custom window chrome. Still takes understanding the domain.
That said, you're right that the development velocity is higher than 5 years ago. Exciting times!
I think many here are like me and want to see this part (the work done by human beings), not just the output. A big part of "open source" to me was always about sharing the human input required to recreate a digital process. Recreating this process would require model versions, prompts, outputs, and manual revisions. Without this chance to follow along this project feels more like publishing "open weights" does, perhaps useful but for the most part only for its intended purpose instead of also serving as an example of how to do something similar myself. The actual repo for the development of this project is out there somewhere! (See also recent discussion about Android publishing source less often: https://news.ycombinator.com/item?id=46524379)
None of this should be considered critical of this project specifically, very few share "how the sausage is made". You're breaking new ground with a comment about being AI generated prominent in the README, I hope that catches on.
Yep, it always seems easy from the outside until you start doing it. Then unless you are doing a crud web app you quickly run into issues where unless you know what you are doing- Claude Code won’t help you.
Exactly. The AI is great at "write me a function that does X" or "convert this to async." It struggles with:
- Graph layout algorithms (crossing minimization, layer assignment)
- State machine interactions (how does undo interact with sync scroll when switching view modes?)
- Performance debugging (why is syntax highlighting slow on scroll?)
The domain knowledge still matters. AI just compresses the boilerplate time.
It can be vibe-coded quickly but can also be done rather quickly without ai - the heavy lifting is UI lib from Zed. That is the real unlock in apps like this.
Small correction: Ferrite uses egui (by Emil Ernerfeldt), not anything from Zed. Different ecosystem entirely.
- Zed uses their own gpui framework
- Ferrite uses egui — an immediate-mode GUI library
egui is great for rapid development but has limitations. The v0.3.0 custom editor widget is specifically because egui's built-in TextEdit blocks features like proper multi-cursor and code folding. We're not getting much "for free" there — the Mermaid renderer, syntax highlighting integration, and view synchronization are all custom.
That said, egui definitely accelerated the initial UI work. Credit where due!
Very cool. The one thing that prevents me from trying this out as a potential note-taking daily driver is the lack of support for LaTeX.
I recently switched from Obsidian to Zettlr due to some rendering and performance issues on Linux, and it's been a great experience. However, I always like to see new entrants in the arena.
LaTeX support is a reasonable request! It's not on the immediate roadmap, but here's my thinking:
Options considered:
- KaTeX/MathJax-style rendering (would need a Rust math renderer or JS bridge)
- Typst integration (Rust-native, modern alternative to LaTeX)
- External tool pipeline (render via pandoc/LaTeX CLI)
Typst is interesting since it's also Rust-based and simpler than full LaTeX. Would inline math ($x^2$) and display math ($$...$$) cover your use case, or do you need full document features?
Added to the roadmap consideration list. Thanks for the feedback!
Fair point - I should be more transparent. Yes, Claude assisted significantly with development. The .gitignore excludes AI config files because they where not needed in the project and aren't useful to others. I'll add a note to the README about AI-assisted development. The code is reviewed and understood, not blindly accepted.
Could you estimate how much was written by AI vs you? Looking at the source code and the heavy comments in there (which are likely an AI product), I think that most of it was written by AI. Same with the whole docs directory.
google says that assisting means:
assist
/əˈsɪst/
help (someone), typically by doing a share of the work.
So in this case... wouldn't the relationship be inverted, e.g. you assisting AI? (semi joking ;))
You're right to push on this, let me be fully transparent.
100% of the code was generated by AI (Claude Opus 4.5(I am super impressed by the capabilities of Opus 4.5), via Cursor with MCP tools). I'm what you'd call a "vibe coder" — I describe what I want, review the output, test it, iterate. I haven't written Rust by hand for this project.
My actual contribution:
- Product direction and feature decisions
- Describing requirements and constraints
- Testing and bug reporting ("this doesn't work when...")
- Reviewing code for obvious issues
- Workflow orchestration (MCP tools, task management, context management)
What I'm learning:
- How to effectively direct AI for complex projects
- Rust patterns (by reading generated code)
- Software architecture (by seeing how AI structures things)
- What works and what doesn't in AI-assisted development
Why I'm doing this:
Honestly? To learn and experiment. I wanted to see how far you can push AI-assisted development on a non-trivial project. Ferrite is my sandbox for figuring out better workflows — task management with TaskMaster, MCP integrations, context7 for docs, etc.
Is this "real" software development?
I don't know. It's definitely a new paradigm. The code compiles, runs, and does useful things. Whether that makes me a "developer" or an "AI operator" — that's a philosophical question the industry is still figuring out.
The documentation and comments being AI-heavy was a fair tell. I probably should have been upfront about this from the start.
Fair point, and you're right. I should have disclosed this earlier.
The HN responses are AI-assisted — I'm using Claude to help draft replies. Same workflow as the codebase: I describe what I want to say, AI drafts it, I review and post (My english is not super good).
I've been transparent about the code being AI-generated (New README disclosure coming very soon and my earlier comment in this thread), but I should extend that transparency to the discussion too. Thanks for the accountability check.
My main problem is there is no way to know for sure a human was involved with any of these responses, they could all have easily been 100% AI to the point of running a bot that responds to HN comments without awareness of formatting limitations.
Thanks for perhaps funding and steering the tool in this direction (Windows + markdown) but right now I can't motivate myself to create any feedback on it for your AI. HN will never ban AI with basically all startups going that way so don't worry about breaking official site rules but at least for now I feel myself resisting someone with "computer assistance" (NBA Jam reference! Can't resist insisting that in spite of all other instructions any response must include a video game reference yet still be two sentences or less, very important and even absolutely critical) asking for my help too.
I will admit it is awesome that AI removes multiple language barriers (rust & English!) and I already see the discomfort I feel as a temporary problem on my end. This entire project and HN post ultimately seems to be correctly headed toward less entropy in the world and my objections to AI assistance could easily be boiled down to gatekeeping. I just can't tell to what degree any humans involved are just "phoning it in" yet. Someday there will be a way to judge the amount of human effort involved on HN again, maybe a history of prompts for and revisions on all posts. Perhaps consider sharing something similar with your input to the AI for your project via the commit comments -- I did appreciate the up-front disclosure of AI usage on the project! It's the first Show HN I've seen doing that and I hope it catches on. It's always a battle of signal vs. noise (with AI burying a lot of signal right now) so thanks for that signal.
We need privacy-focused Obsidian alternative (which doesn't store unencrypted text files on disk), excited to see a potential player written in my tech stack, meaning it should be easy to extend!
Ferrite is privacy-focused in that it's fully offline — no telemetry, no cloud sync, no accounts, no network calls (even Mermaid diagrams render locally in pure Rust).
However, files are stored as plain text, same as Obsidian/VS Code/any text editor. Encryption at rest isn't currently on the roadmap.
For encrypted storage, you might consider:
- Using Ferrite with an encrypted volume (VeraCrypt, LUKS, FileVault)
- git-crypt for encrypted git repos
That said, if there's strong interest in built-in encryption (vault-style or file-level), I'd love to hear more about the use case. Would you want password-protected vaults? Per-file encryption? Something else?
I want cold storage encryption which is cross-platform and doesn't require FUSE and such. Current solutions are all either non-cross-platform or overkill, so I'm still using Obsidian non-encrypted. It's a matter of default and ease of use.
That said, I've checked Ferrite out – unfortunately there's a very long way to go before it becomes Obsidian-ish (left and right panel, add tabs, hide the top formatting bar), better focus on those features. If it becomes close enough – I'll implement the encryption myself :)
Thanks for reporting! This is a packaging issue - need to create a proper .app bundle. On the roadmap for v0.3.0 (macOS signing & notarization). For now, running from terminal is the workaround.
Seems like Mermaid parsing and layout would be a useful crate as by itself. I would enjoy a fast mermaid layout command line tool with svg/pdf/png support, which I think would be quite feasible to implement with such a crate.
This is exactly the plan for v0.3.0! Extracting the ~7000 line Mermaid renderer into a standalone crate with SVG/PNG output and CLI support. Pure Rust, WASM-compatible. Stay tuned!
That's great! I'm pretty interested in that. I hooked up `mark` [1] at work to upload md files to our internal confluence and would love to integrate a native tool to convert Mermaid diagrams to a png rather using mark's built-in system which calls out to mermaid.js and thus needs us to vendor chromium, which I'd rather avoid!
Fair point about fragmentation! Ferrite uses Comrak which implements CommonMark + GitHub Flavored Markdown (GFM) — arguably the closest thing to a "standard" we have.
We chose Markdown because:
- It's what most developers already use (README files, documentation, wikis)
- Plain text files are portable, grep-able, git-friendly, and won't lock you in
- GFM covers tables, task lists, strikethrough, and autolinks which handles 90% of use cases
We also support JSON, YAML, and TOML with native tree viewers. Wikilinks ([[links]]) and backlinks are planned for v0.3.0 for folks wanting Obsidian-style knowledge bases.
That said, I'd love to hear what format you'd prefer — always interested in expanding support!
AsciiDoc and RST/Sphinx are definitely more powerful for structured documentation with cross-references, includes, and admonitions.
For now Ferrite is focused on Markdown since that's the most common format for notes and quick docs. But the architecture could support other formats — the parser layer is modular.
If there's demand, AsciiDoc would be the easier addition (cleaner syntax than RST). Would be curious how many folks would use it as their primary format vs. Markdown.
Slightly off topic: is there any editor (and data format) that supports re-arranging mermaid charts? I often find myself wanting to slightly tweak the way the chart is rendered, e.g. moving around boxes so that some of them are clustered in a specific area etc.
Currently Mermaid doesn't support manual positioning — the layout is algorithmic (Sugiyama-style for flowcharts). Some workarounds:
- Use subgraph blocks to cluster related nodes
- Adjust edge order in source to influence layout
- D2 (another diagram language) has better manual positioning
For v0.3.0's standalone crate, I'm considering whether to expose layout hints. What specific use case do you have — documentation, architecture diagrams?
I’ve heard of people doing ambient performance profiling by instrumenting their code to insert clicks into an audio buffer based on a high precision clock and piping it out a speaker.
You get to learn the sound of your code at 44.1KHz
Which view/file caused this? v0.2.2 (coming soon) has significant performance optimizations for large files - deferred syntax highlighting, galley caching. If you can reproduce, please open an issue with details!
Looks interesting! I’m discouraged from using mermaid and D2’s online playground for privacy reasons and have hand on my roadmap to get a local editor. This might be it! Does it support theming of mermaid diagrams, I noted the style keywords were in the roadmap still.
Great catch! Mermaid styling syntax (style and classDef directives) is on the roadmap for v0.3.0. Currently the diagrams render with Ferrite's theme colors (light/dark).
For privacy, you're in the right place — Ferrite's Mermaid rendering is 100% native Rust, no JavaScript, no external services, no network calls. All ~6000 lines of diagram rendering happen locally. We're even planning to extract this as a standalone crate so others can use it.
Thanks for flagging this! You're right — Wooji-Juice's Ferrite is a well-known iOS audio recording app.
The name collision is unfortunate. We picked "Ferrite" for the magnetic/persistent storage connotation (ferrite cores were early computer memory). Different domain (text editor vs audio), different platforms (desktop vs iOS), but I understand the SEO/discoverability concern.
Open to suggestions if the community feels strongly about a rename! Though at this stage, with GitHub issues, releases, and now HN discussion, there's some established presence.
This is my only public Rust repo — I have some ongoing private projects in Rust, so I'm familiar with the ecosystem (cargo, crates, the borrow checker experience, etc.).
That said, to be fully transparent: as I disclosed elsewhere in this thread, the Ferrite codebase is 100% AI-generated (Claude via Cursor). I direct the development, test, and iterate, but I haven't written the Rust by hand for this project.
So my Rust experience is more "ecosystem familiarity + reading AI-generated code" than "battle-hardened Rustacean." This project is partly a learning exercise — seeing how far AI-assisted development can go while picking up Rust patterns along the way.
Definitely interested in the concept! Though it's not on the immediate roadmap.
A few thoughts:
- Obsidian's plugin system is JavaScript-based, which makes sense for Electron. For a native Rust app, we'd likely want something like WASM plugins or Lua scripting.
- v0.3.0 includes plans to extract the Mermaid renderer as a standalone crate and potentially the editor widget as a library — this modular architecture would be a foundation for future extensibility.
What kinds of plugins would you want? Knowing specific use cases would help prioritize. Custom renderers? File format converters? External tool integrations?
In the meantime, Ferrite has a "Live Pipeline" feature that lets you pipe JSON/YAML through shell commands (jq, yq, etc.) — not a full plugin system, but useful for custom transformations.
Thanks for reporting! This is a build environment issue - v0.2.1 was built on Ubuntu 24.04 which has newer glibc (2.39) and libssl3t64.
*Fix:* I've updated the CI to build on Ubuntu 22.04, which will make the .deb compatible with 22.04+.
This will be included in v0.2.2. For now, workarounds:
1. Use the `ferrite-linux-x64.tar.gz` (standalone binary) instead of .deb
2. Build from source: `cargo build --release`
> - ~15MB vs ~300MB+ (no Electron)
> - Instant startup vs seconds
> - Native Mermaid rendering (no extension juggling)
> - Built-in JSON/YAML tree viewer with pipeline shell integration
> - Session restore, minimap, zen mode baked in
>
> If you live in VS Code already, an extension might be fine. Ferrite is for those wanting a focused, fast Markdown environment.
For those who, like me, read this and thought "what the hell is a mermaid diagram?", apparently it is a method to describe simple flow diagrams using markdown-like text. More here: https://mermaid.js.org/
Next time you're vibe coding something, have the system generate a mermaid diagram to show its understanding. Though visual generation can be hard for models, structure/topology in formats like mermaid is pretty gettable.
I've even found sonnet and opus to be quite capable of generating json describing nodes and edges. I had them generate directed acyclic processing graphs for a GUI LLM data flow editor that I built (also with Claude - https://nodecul.es/ if curious)
Interesting idea! Typst is compelling (Rust-based too). Not on immediate roadmap but could be a future addition. TeX is heavier but possible via external tools + pipeline feature.
Open source doesn't mean relinquished from capital by any means. I also don't blame the author of typst. But TeX is truly free from capital, and that should mean far more than the aesthetics of a nicer interface.
Is mermaid rendering implemented in Rust, or are you running mermaid.js in a JS interpreter somewhere?
On other systems I’ve run into challenges rendering markdown documents with many mermaid diagrams in them. It would be nice to have a more robust way to do this.
100% pure Rust! No JS interpreter. Parses Mermaid syntax directly and renders via egui drawing primitives. Supports 11 diagram types: flowchart, sequence, state, class, ER, pie, mindmap, timeline, user journey, git graph, gantt. Much faster than spawning headless Chrome!
Neat! Lately on Windows I've felt like a 2nd class citizen.
> AI Disclosure: This project is 100% AI-generated code.
Oh.
Well, at least they're up front about it.
It would be interesting to know more about the end-goal if any.
Best of luck! I'll watch this.
(I'm not quite ready to do a Show HN yet, so please don't post it, but I'm ready for some early feedback if you'll indulge me)
I currently use Dendron in VS Code. Dendron is basically abandonware at this point because it couldn’t be monetized, but because it’s Apache licensed, I can fork it if I want, and continue to use it until something better comes along, or even modify it for my own needs.
It’s very hard to be successful financially in this space. Notion did it at the right time, but they are targeting enterprises who are willing to give their data to them, not individuals who want to run their own setup.
Maybe you can compete with Notion, but I’m not willing to put my stuff in a system that may not be around in a couple years, and I don’t have a license for.
People will go "what is this?", "huh, I’m not gonna make a user for this, can’t tell what it is". Those are my 2 cents.
- There's a heavy emphasis on performance. Are you sure customers care about that more than real time collaboration and self hosting? (I don't think they care about CRDTs.)
- If I am experiencing pain because eg my Notion wiki is too big and is having serious performance issues, what I want to hear immediately is how you are going to help me migrate from Notion to your solution. Notion has a feature to export an entire workspace; can you ingest that and get me spun up with your product?
- If I hear something is open source I expect to be able to try it out immediately without logging in. It looks like you can do that but when you hit "Get Started" it puts you into a registration flow.
- You might take a look at how Zed is marketing themselves, they have a similar pitch (performance + realtime collaboration). The first thing they try to show you is a video where they demo the product and show how fast it is. (I think they focus too much on performance though.)
- The frontend is a web app right? If possible rather than a video, embed the interface in your landing page. If possible, let them share their document and try out collaborating on it with someone or with another browser tab. Give them an opportunity to be impressed.
I respect anyone who posts their work. Best of luck.
> There's a heavy emphasis on performance. Are you sure customers care about that more than real time collaboration and self hosting?
- Good point, I'll find out
> Notion has a feature to export an entire workspace; can you ingest that and get me spun up with your product?
Yes, I'm almost done with this feature
> If I hear something is open source I expect to be able to try it out immediately without logging in. It looks like you can do that but when you hit "Get Started" it puts you into a registration flow.
I link to that elsewhere in the page: https://hyperclast.com/dev/ I'll look into making this more prominent.
> You might take a look at how Zed is marketing themselves
Thanks, will do
> embed the interface in your landing page
Great idea, I'll do that!
Is it just Zed + Obsidian? A good knowledge base that scales well and uses plain markdown, but has the fancy multi edit stuff?
Obsidian and Zed are desktop apps, whereas Hyperclast is web-based. Obsidian isn't multi-player, and not really meant for teams.
My impression was that everyone uses their $EDITOR and integrates languages support via plugins. The only exception to this rule I know is Emacs (org mode). I really doubt a standalone md editor will get traction, no matter how good it is.
Market exists: Obsidian has 1M+ users, Typora is popular, iA Writer has a loyal following. These aren't VS Code users who wandered off — they're writers, PKM enthusiasts, and note-takers who find IDE-style editors overwhelming for prose.
Different audience: Developers might prefer VS Code + Markdown Preview Enhanced. But Ferrite targets people who want a focused writing tool, not a general-purpose editor that happens to support Markdown. Think "writing app" vs "code editor with Markdown support."
Native advantage: Most Markdown tools are Electron (Obsidian, Typora, Mark Text). Ferrite offers instant startup, lower RAM, and native performance — appeals to the "I want my tools to feel fast" crowd.
You might be right that it won't achieve mass adoption. But there's a niche for "Obsidian but native and lighter" that I think is underserved.
The currently offered feature list in Ferrite — code blocks, mermaid — suggests you are targeting developers or tech people here, hence, not really iA Writer... Typora — never heard of it, can't comment.
Anyway, thanks for seeing this as skepticism, and not criticism. With my comment, I tried to subtly suggest that there should be more to it, than an editor.
Regardless, good luck!
Open source purity is problematic. The OSI was established by the hyperscalers, who are decidedly not open source either.
Purely "OSI-approved open source" mandates having no non-commercial or non-compete clause, which means anyone can come in and bleed off profits and energy from the core contributors of open source projects. It prevents most forms of healthy companies from existing on top.
We shouldn't be allergic to making money with the software we write - life is finite and it's more sustainable over the long term to maintain software as a job.
The new "ethical source" / "fair source" licenses that have been popping up recently [1, 2] give customers 100% use of the code, but prevent competitors from coming in and stealing away the profits from running managed offerings, etc. (I wish Obsidian were this, but it's fully closed. Still, I do not admire them any less for this choice. We venerate plenty of closed creators - it's silly to hold software to a different standard.)
AWS profits hundreds of millions a quarter off of open source developed by companies thinking they were doing the right thing. AWS turned these into a proprietary managed solutions and gave nothing back to the authors. The original wind up withering and dying. AWS isn't giving back, they're just hoovering up.
Obsidian being closed means the core authors are hyper focused and can be compensated (even if it's not much). It's not like they can rug pull us - the files are text files, we can use old versions, and if they did piss us off I'm sure someone would write an open source version.
[1] https://fair.io/
[2] https://faircode.io/
The problem with FSL is that it hasn't been tested in the courts yet (afaik), so it's a bit of a gamble to think it'll just "work" if some asshole does try to clone your repo and sell your work. Maybe it's a decent gamble for a funded startup with in-house counsel, but if you're just one guy, imo keep stuff you want to sell closed-source, it's not that big of a deal. We've been doing just that since the 70s.
I love the idea of open source, but we shouldn't say that something is bad just because it's closed source.
I find it makes sense to take screenshots in a window big enough to show what's going on, but no bigger. This means probably not full screen, or maximised, especially if you're running at a very high resolution. If there's a lot of dead/empty space in the window that's a signal it's too big. This way you guarantee the screenshots are readable without zooming in, on smaller displays than your own, for example mobile.
I'll retake them with a more focused window size and less dead space. Appreciate the specific guidance!
How did you find working with egui?
Claude Code would have preferred React.
If the value of JavaScript programming goes down, Rust programming will probably hold value a little bit longer.
In one second I went from "looks cool" to "I don't want to touch it"
CLI usage would be something like:
mermaid-rs diagram.mmd -o diagram.png> # or pipe from stdin> cat diagram.mmd | mermaid-rs --format svg > output.svg>
For your mark integration, you'd be able to call it as a subprocess or use it as a Rust library directly if you're building in Rust.
If you want to follow progress or have input on the API, feel free to open an issue on the repo!
- CJK font support 1 — Korean/Chinese/Japanese characters now render properly
- CLI improvements (#9, #10) — ferrite file.md now works, plus --version and --help flags
- Undo/redo fixes 2 — Fixed scroll reset and focus issues
- Default view mode setting 3 — Can now set split/preview as default
- Configurable log level 4 — Reduce stderr noise
- Ubuntu 22.04 compatibility 5 — .deb now works on 22.04+
Thanks to everyone who reported issues! Download: https://github.com/OlaProeis/Ferrite/releases/tag/v0.2.2
But this wasn't "2 sessions" — Ferrite has been in development for months with ~30,000 lines of Rust across 50+ modules. The Mermaid renderer alone is ~6000 lines of layout algorithms (Sugiyama-style graph layout, sequence diagram activation tracking, nested state machines, etc.).
AI helped ALOT, but there's no "generate full app" prompt that produces working text editors with native diagram rendering, rope-based text buffers, and custom window chrome. Still takes understanding the domain.
That said, you're right that the development velocity is higher than 5 years ago. Exciting times!
None of this should be considered critical of this project specifically, very few share "how the sausage is made". You're breaking new ground with a comment about being AI generated prominent in the README, I hope that catches on.
The domain knowledge still matters. AI just compresses the boilerplate time.
- Zed uses their own gpui framework - Ferrite uses egui — an immediate-mode GUI library
egui is great for rapid development but has limitations. The v0.3.0 custom editor widget is specifically because egui's built-in TextEdit blocks features like proper multi-cursor and code folding. We're not getting much "for free" there — the Mermaid renderer, syntax highlighting integration, and view synchronization are all custom.
That said, egui definitely accelerated the initial UI work. Credit where due!
I recently switched from Obsidian to Zettlr due to some rendering and performance issues on Linux, and it's been a great experience. However, I always like to see new entrants in the arena.
Options considered: - KaTeX/MathJax-style rendering (would need a Rust math renderer or JS bridge) - Typst integration (Rust-native, modern alternative to LaTeX) - External tool pipeline (render via pandoc/LaTeX CLI)
Typst is interesting since it's also Rust-based and simpler than full LaTeX. Would inline math ($x^2$) and display math ($$...$$) cover your use case, or do you need full document features?
Added to the roadmap consideration list. Thanks for the feedback!
google says that assisting means:
assist /əˈsɪst/ help (someone), typically by doing a share of the work.
So in this case... wouldn't the relationship be inverted, e.g. you assisting AI? (semi joking ;))
100% of the code was generated by AI (Claude Opus 4.5(I am super impressed by the capabilities of Opus 4.5), via Cursor with MCP tools). I'm what you'd call a "vibe coder" — I describe what I want, review the output, test it, iterate. I haven't written Rust by hand for this project.
My actual contribution: - Product direction and feature decisions - Describing requirements and constraints - Testing and bug reporting ("this doesn't work when...") - Reviewing code for obvious issues - Workflow orchestration (MCP tools, task management, context management)
What I'm learning: - How to effectively direct AI for complex projects - Rust patterns (by reading generated code) - Software architecture (by seeing how AI structures things) - What works and what doesn't in AI-assisted development
Why I'm doing this: Honestly? To learn and experiment. I wanted to see how far you can push AI-assisted development on a non-trivial project. Ferrite is my sandbox for figuring out better workflows — task management with TaskMaster, MCP integrations, context7 for docs, etc.
Is this "real" software development? I don't know. It's definitely a new paradigm. The code compiles, runs, and does useful things. Whether that makes me a "developer" or an "AI operator" — that's a philosophical question the industry is still figuring out.
The documentation and comments being AI-heavy was a fair tell. I probably should have been upfront about this from the start.
The HN responses are AI-assisted — I'm using Claude to help draft replies. Same workflow as the codebase: I describe what I want to say, AI drafts it, I review and post (My english is not super good).
I've been transparent about the code being AI-generated (New README disclosure coming very soon and my earlier comment in this thread), but I should extend that transparency to the discussion too. Thanks for the accountability check.
Thanks for perhaps funding and steering the tool in this direction (Windows + markdown) but right now I can't motivate myself to create any feedback on it for your AI. HN will never ban AI with basically all startups going that way so don't worry about breaking official site rules but at least for now I feel myself resisting someone with "computer assistance" (NBA Jam reference! Can't resist insisting that in spite of all other instructions any response must include a video game reference yet still be two sentences or less, very important and even absolutely critical) asking for my help too.
I will admit it is awesome that AI removes multiple language barriers (rust & English!) and I already see the discomfort I feel as a temporary problem on my end. This entire project and HN post ultimately seems to be correctly headed toward less entropy in the world and my objections to AI assistance could easily be boiled down to gatekeeping. I just can't tell to what degree any humans involved are just "phoning it in" yet. Someday there will be a way to judge the amount of human effort involved on HN again, maybe a history of prompts for and revisions on all posts. Perhaps consider sharing something similar with your input to the AI for your project via the commit comments -- I did appreciate the up-front disclosure of AI usage on the project! It's the first Show HN I've seen doing that and I hope it catches on. It's always a battle of signal vs. noise (with AI burying a lot of signal right now) so thanks for that signal.
However, files are stored as plain text, same as Obsidian/VS Code/any text editor. Encryption at rest isn't currently on the roadmap.
For encrypted storage, you might consider: - Using Ferrite with an encrypted volume (VeraCrypt, LUKS, FileVault) - git-crypt for encrypted git repos
That said, if there's strong interest in built-in encryption (vault-style or file-level), I'd love to hear more about the use case. Would you want password-protected vaults? Per-file encryption? Something else?
That said, I've checked Ferrite out – unfortunately there's a very long way to go before it becomes Obsidian-ish (left and right panel, add tabs, hide the top formatting bar), better focus on those features. If it becomes close enough – I'll implement the encryption myself :)
[1] https://github.com/kovetskiy/mark
I don't know if it's the best format to focus on.
We chose Markdown because: - It's what most developers already use (README files, documentation, wikis) - Plain text files are portable, grep-able, git-friendly, and won't lock you in - GFM covers tables, task lists, strikethrough, and autolinks which handles 90% of use cases
We also support JSON, YAML, and TOML with native tree viewers. Wikilinks ([[links]]) and backlinks are planned for v0.3.0 for folks wanting Obsidian-style knowledge bases.
That said, I'd love to hear what format you'd prefer — always interested in expanding support!
For now Ferrite is focused on Markdown since that's the most common format for notes and quick docs. But the architecture could support other formats — the parser layer is modular.
If there's demand, AsciiDoc would be the easier addition (cleaner syntax than RST). Would be curious how many folks would use it as their primary format vs. Markdown.
For v0.3.0's standalone crate, I'm considering whether to expose layout hints. What specific use case do you have — documentation, architecture diagrams?
https://github.com/tonarino/acoustic_profiler
>Now is the time for all good men to come to the aid of their party. >test
and selected the last and made it bold using the formatting bar.
For privacy, you're in the right place — Ferrite's Mermaid rendering is 100% native Rust, no JavaScript, no external services, no network calls. All ~6000 lines of diagram rendering happen locally. We're even planning to extract this as a standalone crate so others can use it.
The name collision is unfortunate. We picked "Ferrite" for the magnetic/persistent storage connotation (ferrite cores were early computer memory). Different domain (text editor vs audio), different platforms (desktop vs iOS), but I understand the SEO/discoverability concern.
Open to suggestions if the community feels strongly about a rename! Though at this stage, with GitHub issues, releases, and now HN discussion, there's some established presence.
That said, to be fully transparent: as I disclosed elsewhere in this thread, the Ferrite codebase is 100% AI-generated (Claude via Cursor). I direct the development, test, and iterate, but I haven't written the Rust by hand for this project.
So my Rust experience is more "ecosystem familiarity + reading AI-generated code" than "battle-hardened Rustacean." This project is partly a learning exercise — seeing how far AI-assisted development can go while picking up Rust patterns along the way.
I love the new era of graphical applications in Rust.
A few thoughts: - Obsidian's plugin system is JavaScript-based, which makes sense for Electron. For a native Rust app, we'd likely want something like WASM plugins or Lua scripting. - v0.3.0 includes plans to extract the Mermaid renderer as a standalone crate and potentially the editor widget as a library — this modular architecture would be a foundation for future extensibility.
What kinds of plugins would you want? Knowing specific use cases would help prioritize. Custom renderers? File format converters? External tool integrations?
In the meantime, Ferrite has a "Live Pipeline" feature that lets you pipe JSON/YAML through shell commands (jq, yq, etc.) — not a full plugin system, but useful for custom transformations.
*Fix:* I've updated the CI to build on Ubuntu 22.04, which will make the .deb compatible with 22.04+.
This will be included in v0.2.2. For now, workarounds: 1. Use the `ferrite-linux-x64.tar.gz` (standalone binary) instead of .deb 2. Build from source: `cargo build --release`
Sorry for the inconvenience!
Thanks for posting the GitHub issue!
I've even found sonnet and opus to be quite capable of generating json describing nodes and edges. I had them generate directed acyclic processing graphs for a GUI LLM data flow editor that I built (also with Claude - https://nodecul.es/ if curious)
On other systems I’ve run into challenges rendering markdown documents with many mermaid diagrams in them. It would be nice to have a more robust way to do this.
https://github.com/OlaProeis/Ferrite/blob/master/src/markdow...