Went to watch the Youtube intro [1] for Penguin Mail and was greeted by the most egregious scam video. Didn't click the ad, but simply went to the domain (usbest1.com) and it's just chock full of AI generated content hoping to scam you out of your money. Going through the checkout flow, it even asks you if you want to tip your driver (presumably for delivery).
Sorry for the off-topic, but Google (and all ad networks) should be legally required to vet their advertisers. That's gotta be the worst I've seen.
Amazed anyone here isnt using an ad blocker for youtube tbh, it feel like the first thing you'd want to avoid the mess.
Ublock is still a thing. Or use Brave as thats got baked in ad blocking which can be combined with ublock to make it even more effective. Then adguard or pihole on your local network so no analytics/tracking data is still being pushed out.
Interestingly, one of Poland’s wealthiest individuals (who has been the target of scams ranging from malicious to macabre) is now engaged in an openly declared war with Meta[1][2]. There is an entire foundation dedicated to tracking ads and identifying scams; they calculate the profits generated from these schemes and lobby for regulations that would mandate the vetting of advertisers, recently on an EU level.
I went through a phase of reporting them (to YouTube, there's a button while playing to do so) and did get automated reply that a lot were removed. Some took several reports, others they continued to insist didn't violate their policies, but they will remove them.
YouTube has some seriously scammy ads. To an extent that these ads have motivated me to buy premium for my parents as they tend to fall into the trap of fake dialog boxes shown in ads warning a PDF app needing an update.
The thing with all these vibe-coded email clients is: how well do they sanitize the full-content view of HTML emails?
Thunderbird and the like go to great lengths to make viewing those emails safe, e.g., by preventing inline JavaScript and all kinds of tracking techniques (pixels, CSS references, etc.). Without having checked, do these vibe-coded clients just render the HTML in a WebView, or do they have similar protections in place?
This is actually a very tricky problem. We solve this by never rendering mails in trusted origin. It's basically sanitisation + isolated sandboxed iframe for defence in depth + standard security stuff like CSP headers. When I was looking into this, I was surprised how many popular mail clients relay primarily on sanitization and render mails within trusted origin (ex: marco, superhuman).
Other interesting problem to solve for mail clients is the editor. You should be able to forward/edit untrusted html content. Source: I develop tecotype.com for mac/linux/windows.
It seems the <sarcasm> tag was missing, I assume,
but as shown by many companies and open source projects, agents with the right prompts are much better at finding security problems than 99% of developers (for various reasons).
Many devoted fans of AI like comparing LLMs to compilers for higher-level programming languages.
They argue that just as programmers once wrote ASM and C, the arrival of Java and Python meant you no longer had to be a neckbearded autistic kid to write software that doesn't leak memory or segfault.
Arguments like this show me you’ve never done software engineering seriously.
Despite compilers (written by some of the smartest engineers on Earth) trying hard to prevent developers from writing shitty software, those new age "engineers" still managed to ship software that crashes, leaks, and generally sucks.
I'm not saying this email client is not secure (although I'd bet 5$, it is). But it seems beyond silly to assume software is automatically safe just because a new piece of tech can potentially find vulnerabilities. Intent still matters. And the 99% would never have the intent to write secure software.
That's a very nice looking app - I like the look of it it gives me Mail.app vibes.
A mail client is always a tough thing to get right so I'm glad you're working on one
I love then idea that as the total cost (time/money/attention) of personalizing software continues to come down, more people will be able to encourage a diaspora of options, tailored to each niche.
You need a secure, supported mail client first and foremost. Better looks definitely would help Linux adaption, but safety and functionality takes precedence in this case
I don't see any good reason to assume that these apps are less secure just because AI is used in the making of them. Bad developers and bad development practices existed long before LLMs.
I agree. The other day I reinstalled some old gaming pc of mine with the latest bazzite. Then I discovered it had lots of old photos on it, but in many many backup folders.
So I had Claude whip up a super fast image browser that had some features I really like from a Mac app I use. It looks better than 90% of Linux apps, is super fast and fits me like a glove.
Of course some of these mail apps will be abandoned, but they also provide an opportunity to make things more attractive on the Linux side and win some users.
Visually it resembles Mail.app, but if it’s going for a similar level of functionality I think modern GNOME’s hamburger-and-three-dots-buttons approach to menus is going to become a problem.
This is one of those types of app that really benefits from a proper menubar with several top-level menus so the user isn’t stuck spelunking submenu tunnels.
Because "Find me that mail from last year from Peter or Pat? that I never replied to" is easier for most people then setting a bunch of filters and regex.
Quick grammar correction can also be nice. Not super important, but nice.
But these MUST have a local option first and foremost, this is something that any local model can do, even the smallest ones.
But even before that there would need to be a set of APIs from the DE, imho. I think that GNOME is going into this direction with their AI center, i whish KDE would start looking into that too
I've been using Aerion (https://aerion.3df.io) for a few months now, it's another mail+calendar+contacts app for Linux that is largely vibecoded. I haven't tested Penguin thoroughly yet, but I think Aerion is more feature rich, and has served me well over the months.
Why am I not using Thunderbird? Aerion's UI is much better, and most importantly snappier, especially on my older laptop.
Interesting that they call out Fastmail as a supported provider, but don't support JMAP. Seems like that would have been an easy stretch goal for Claude, and a good way to stand out amongst the competition.
I wouldn't say it's something that can be vibecoded in a month or two. Proper design and implementation take time if you want to handle accounts with hundreds of thousands of messages seamlessly.
These days with the convenience of having LLMs at developer’s disposal, it feels unjustified to use electron like apps.. whenever I see “one app all/multi-platform”, I feel discouraged to give it a genuine try.
I tried a bunch of email clients and then crawled back to mail.app + Calendar.app on MacOS simply for how native it feels.
A project with the complexity of an email client cannot be considered stable in a month. There are bugs that will only show up after a year. It has a calendar after all.
> The UI looks pretty good too. It's really not hard to beat thunderbird.
I don't understand this mentality. It's an email client. There's not a lot needed UI-wise to make a functioning interface. Display a list of emails, show the full message when you open it, etc. I don't see anything functionally different between this email app and basically every other one I've ever used.
You've either got insane standards or are using email way different than I.
Can’t speak for others but for me Thunderbird’s UI feels distracting and almost grating in some ways. It’s technically functional but doesn’t feel right.
I know things like that don’t bother some folks and that’s fine, but others like myself can’t turn the visual parts of their brains off and so some software feels like nails on a chalkboard to use.
> It’s technically functional but doesn’t feel right.
That might be the most accurate description of Thunderbird that I've come across. It's hard to pinpoint exactly what's wrong with Thunderbird, despite me using it every day. Somehow it feels like the emails aren't the important bit of the UI, but I also couldn't tell you what I'm looking at instead.
> There's not a lot needed UI-wise to make a functioning interface
I remember waiting for ~15 years until Thunderbird finally implemented rich content support in the e-mail list (needed for a proper 3-column layout, as otherwise the e-mail list, being a simple table, wouldn't fit on the widest screens available).
So, yeah... Not a lot, but Thunderbird has been missing the bar for over a decade.
But from a normal user's standpoint? I don't care. If my email client needs more than a single thread, something's fucked.
I randomly bounce between Thunderbird and KMail and honestly I can't tell the difference except cosmetically. Both are snappy. Neither crash. Both send and receive emails just fine.
I know parent is probably talking about a different type of threading, but Thunderbird's lack of process threading means the UI just gets stuck when it's syncing a large numbers of emails, which can happen if your subscribed to a lot of mailing lists, etc.
There is a case to be made to always start at 1.0, especially libraries with public API’s. The reason is that if you use semantic versioning then below 1.0 you can’t properly declare a new release to be a feature or bugfix release (for example, 0.4 to 0.5. Is this bugfix or feature?). Above 1.0.0 this is easy (for example 1.1.0 vs 1.0.1).
In libraries, 0.x is for when you’re still figuring the API out. Basically you’re saying that no API is stable at this point, and you’re not doing semver yet, and anything can change.
Naturally, the users should treat those libraries as beta versions and only use them in production at their own risk.
> you can’t properly declare a new release to be a feature or bugfix
Nice UI but untrusted email messages is something I'd rather open in [an uglier app with decades long reputation for being reasonably secure](https://www.thunderbird.net/en-US/)
Would not be a problem with me, download source, let the agent search for backdoors and security problems for some hours and it would be fine for me. I would need to do the same with thunderbird if I'd need to trust an UI frontend to my email.
Not sure if this is gaslighting, an absurdist joke, or we are going through a collective psychosis.
KindOfCoders, AI is a very useful tool, but nothing works like the way you're explaining. I am kind of finding it hard to articulate the problem, because the assertion is just a bit wild. So I will do my best.
Thunderbird has been in production use for over 20 years, many many smart people have put their time into it to make sure it works correctly, yes, it is a bit clunky, it has a lot of craft, but you're not going to replace that much labour with running an AI agent for a few hours; despite the AIs usefulness in finding bugs and vulnerabilities. Not yet at least.
You are correct; however, agents can and do find terrible bugs in some of the most popular software in the world made by some of the most competent engineers on the planet.
I assume what KingOfCoders suggested is to run an agent almost like an antivirus which I guess sounds somewhat interesting but you'd have to re-run it on every update and probably spend many more hours+tokens for such a system to work properly.
But perhaps once inference gets truly cheap repository maintainers should include AI security checks for new software as part of their pipeline? I think that would make sense. Especially for repos that are frequently pwned like npm!
How so? Any specific critique? Ive been using Thunderbird tot décades and while e-mail seems to have some inherent clunkiness I find it overall quite fine.
I think this looks good, but I'm not on a Mac anymore.
For people that want a terminal client with tabs, integrated html viewer(including images), markdown compose, MCP, with unified inbox, I built notmutt.
“Email client” is an interesting puzzle. Email involves multiple activities:
Viewing: I have a message or messages stored on disk in files or mboxes or a maildir. How do I render them? Example use case is evidence in a law suit or, more likely, emails I’ve put in a shared drive to help me and my partner argue with our landlord. Yes, I could just have a Gmail folder for these but I don’t want to have one foot in a file-filesystem and one in a mail-filesystem. I want “lease_2024.pdf” and “20240401_deposit_confirmation_002.eml” to both be files. But then of course we have…
Filesystem interface: Emails have important attachments with MIME file names. Or perhaps a more keen observation is that some named files have emails as attachments! I’d like to symlink into a virtual FS that keys files off a UUID / Message-ID, and be able to mount that virtual FS from any mail store I have available (Gmail API, JMAP, IMAP, .tar archive, mbox, maildir, or .eml files.) Most conversations I have with people don’t involve files, and most files don’t involve conversations, but when they do I want them to feel natural and not some kind of special category of filey-messagey thing that doesn’t play well with plain files or, less commonly, plain messages.
Composition and sending: Creating a new message involves making a new file and then using that file to drive a sending API. I want to compose a beautiful email using a tool that is good at that job. Maybe I’m using vim to draft it in markdown with macros and tools to machine build certain complex parts? I also need some mime aware tooling — sometimes — to help prepare a bunch of attachments that may or may not have names related to their names on disk! I want all of this to be scriptable, on the occasions when it’s complex (30 part message for a client’s loan application!) and easy when it’s ad hoc (to: mom, see cat.jpg).
Search: email is notoriously the extension of and replacement for my long term memory. I have gathered message history from many sources and accounts over the years. I need to manage these into packed archives for the old ones, editable archives for the current ones, promote the former to the latter, and arrange new messages into the system. This feels like another separate tool.
I’m sure there are other potential tools lurking here as well. Separate rendering, composition, sending, receiving, searching, and archiving, feels like the goal. Most new email apps I see try to implement all of the above in one new package when, really, it would be great to see innovation on each branch of email as separate tools.
My PDF viewer doesn’t need to innovate on file browsing! My SVG drawing tool doesn’t need to innovate on mounting remote filesystems via SSH!
> - can’t choose a sender email?! This is major dealbreaker, if you have many aliases or domains.
Ugh, this drives me nuts. It's the same reason why I stopped using my Supernote Nomad for email.
I use iCloud+ email with a custom domain. With that arrangement you authenticate to Apple's servers with your iCloud "primary address", like someone.somename@icloud.com. Then you can send and receive email with your custom domain email address, which is the one you're actually using. But not with Supernote, or apparently Penguin Mail. My Nomad will only send mail with my @icloud.com address, even though that's only used for authentication.
So yeah, I just stopped using it altogether. That's not my address. It's not how people reach me. It's literally just a username.
This is interesting, I am doing an experimental effort with similar goals. What is your approach to testing? The bug surface of something like this is quite large. Also, do you have a plan for sandboxing?
I would love an email client that automatically sorts my mails via a local LLM.
As long as it tries to be the "I" in "AI" and do something useful like automatically detect conversations with my accountant, per project and not per sender mails including somehow labeling mails referring to multiple projects with multiple labels etc.
Second that. I can’t really think of a better application for LLMs than this. In fact, this might be the only good application.
Drafting emails with LLMs is... ugh.
And assistant chat? Really?
Okay, two things might be great: translations (preferably local, e.g. using Project Bergamot) and message summary (in case somebody does send an unedited blob of LLM slop to me and social dynamics don’t let me send it to spam).
Then my 'amail' client seems to fit the bill.. (OSS, rust, I use it on Linux and haven't tested elsewhere) ..not really suggesting it – it's pretty barebones and just whatever I've needed, but you're welcome to it obviously! Especially if you have issues/PRs to spur me on to features I've wanted too...
Penguin Mail is a mail and calendar app for Linux, written in Rust with GTK 4 and libadwaita. It's designed to be a lightweight, privacy-first alternative to Thunderbird or GNOME's built-in email.
What it does:
• Aggregates Gmail, Microsoft (Outlook/Hotmail/Microsoft 365), Fastmail, iCloud, Yahoo, and any IMAP/POP3 account into one inbox
• Calendar support for every account (Google Calendar, CalDAV)
• Contacts with CardDAV integration
• OpenPGP and S/MIME signing/encryption — keys never leave your machine
• Gmail-style categories bar (Primary, Updates, Promotions, Social)
• Send Later, Undo Send, templates, Markdown compose
• A built-in AI assistant that can summarize threads, draft replies, manage rules, and search calendars. It runs on a local model (Ollama/LM Studio), an Anthropic API key, or Claude Code — you pick per job
• Remote content blocked twice (WebKit filter + CSP), JS off by default
• Runs in the tray at ~55 MB
It's GPL-3.0 and available as .deb, .rpm, Arch package, snap, tarball, and Flatpak. Not on Flathub yet — that's next.
We built it because existing Linux email clients felt bloated or didn't respect privacy. The AI assistant is a nice-to-have but not required — the app works great without it.
Feedback welcome: install experience, missing features, anything confusing. Happy to answer questions about the Rust/GTK4 architecture too.
Yet another GTK4/libadwaita, no thank you. I mean, if you're going to use Rust and vibe code thing, why not use a more sensible, Rust-native UI like egui?
Maybe I'm not the target here but an embedded chat ai window is a huge turn off to me. Having a CLI I can give my own agent to access/query the data, if I choose to do that, would have been preferred but if the target was my mother, then a chat window is probably a better option.
Went to watch the Youtube intro [1] for Penguin Mail and was greeted by the most egregious scam video. Didn't click the ad, but simply went to the domain (usbest1.com) and it's just chock full of AI generated content hoping to scam you out of your money. Going through the checkout flow, it even asks you if you want to tip your driver (presumably for delivery).
Sorry for the off-topic, but Google (and all ad networks) should be legally required to vet their advertisers. That's gotta be the worst I've seen.
[1]: https://www.youtube.com/watch?v=0PyJCsw1FSE
Amazed anyone here isnt using an ad blocker for youtube tbh, it feel like the first thing you'd want to avoid the mess.
Ublock is still a thing. Or use Brave as thats got baked in ad blocking which can be combined with ublock to make it even more effective. Then adguard or pihole on your local network so no analytics/tracking data is still being pushed out.
Interestingly, one of Poland’s wealthiest individuals (who has been the target of scams ranging from malicious to macabre) is now engaged in an openly declared war with Meta[1][2]. There is an entire foundation dedicated to tracking ads and identifying scams; they calculate the profits generated from these schemes and lobby for regulations that would mandate the vetting of advertisers, recently on an EU level.
[^1]: https://www.bloomberg.com/news/articles/2026-08-27/-locker-k... [^2]: https://www.aa.com.tr/en/europe/polish-minister-accuses-meta...
Protect yourself and your family. Use an adblocker.
I went through a phase of reporting them (to YouTube, there's a button while playing to do so) and did get automated reply that a lot were removed. Some took several reports, others they continued to insist didn't violate their policies, but they will remove them.
YouTube has some seriously scammy ads. To an extent that these ads have motivated me to buy premium for my parents as they tend to fall into the trap of fake dialog boxes shown in ads warning a PDF app needing an update.
Especially now these scam videos can be made on an industrial scale.
Ublock origin
The thing with all these vibe-coded email clients is: how well do they sanitize the full-content view of HTML emails? Thunderbird and the like go to great lengths to make viewing those emails safe, e.g., by preventing inline JavaScript and all kinds of tracking techniques (pixels, CSS references, etc.). Without having checked, do these vibe-coded clients just render the HTML in a WebView, or do they have similar protections in place?
This is actually a very tricky problem. We solve this by never rendering mails in trusted origin. It's basically sanitisation + isolated sandboxed iframe for defence in depth + standard security stuff like CSP headers. When I was looking into this, I was surprised how many popular mail clients relay primarily on sanitization and render mails within trusted origin (ex: marco, superhuman).
Other interesting problem to solve for mail clients is the editor. You should be able to forward/edit untrusted html content. Source: I develop tecotype.com for mac/linux/windows.
"claude, make no mistakes" is your best protection
It seems the <sarcasm> tag was missing, I assume, but as shown by many companies and open source projects, agents with the right prompts are much better at finding security problems than 99% of developers (for various reasons).
Many devoted fans of AI like comparing LLMs to compilers for higher-level programming languages.
They argue that just as programmers once wrote ASM and C, the arrival of Java and Python meant you no longer had to be a neckbearded autistic kid to write software that doesn't leak memory or segfault.
Arguments like this show me you’ve never done software engineering seriously.
Despite compilers (written by some of the smartest engineers on Earth) trying hard to prevent developers from writing shitty software, those new age "engineers" still managed to ship software that crashes, leaks, and generally sucks.
I'm not saying this email client is not secure (although I'd bet 5$, it is). But it seems beyond silly to assume software is automatically safe just because a new piece of tech can potentially find vulnerabilities. Intent still matters. And the 99% would never have the intent to write secure software.
[delayed]
> but as shown by many companies
… but as shown by many shovel sellers… their shovel…
Here I fixed it.
Sometimes it's easier to start with doing the right thing than to just try to fix problems later.
That's a very nice looking app - I like the look of it it gives me Mail.app vibes.
A mail client is always a tough thing to get right so I'm glad you're working on one
I love then idea that as the total cost (time/money/attention) of personalizing software continues to come down, more people will be able to encourage a diaspora of options, tailored to each niche.
There’s been a flurry of vibe coded mail apps for Linux in the last couple of months. Hylki is another one. They all look basically the same too.
They will probably all be abandoned within the year.
They all look and behave better than the old known Linux mail clients.
Linux really need applications more polished, that’s what they are bringing to the table.
You need a secure, supported mail client first and foremost. Better looks definitely would help Linux adaption, but safety and functionality takes precedence in this case
I don't see any good reason to assume that these apps are less secure just because AI is used in the making of them. Bad developers and bad development practices existed long before LLMs.
I agree. The other day I reinstalled some old gaming pc of mine with the latest bazzite. Then I discovered it had lots of old photos on it, but in many many backup folders.
So I had Claude whip up a super fast image browser that had some features I really like from a Mac app I use. It looks better than 90% of Linux apps, is super fast and fits me like a glove.
Of course some of these mail apps will be abandoned, but they also provide an opportunity to make things more attractive on the Linux side and win some users.
Would you like to share the app?
Visually it resembles Mail.app, but if it’s going for a similar level of functionality I think modern GNOME’s hamburger-and-three-dots-buttons approach to menus is going to become a problem.
This is one of those types of app that really benefits from a proper menubar with several top-level menus so the user isn’t stuck spelunking submenu tunnels.
I really hope we’ll see a KDE counterpart (and not in Kirigami; a proper one).
I'm not anti-AI but I really don't see why they felt the need to stuff an AI chatbox in my emails, other that "because they could".
Because "Find me that mail from last year from Peter or Pat? that I never replied to" is easier for most people then setting a bunch of filters and regex.
Quick grammar correction can also be nice. Not super important, but nice.
> find stuff
yeah, that too.
But these MUST have a local option first and foremost, this is something that any local model can do, even the smallest ones.
But even before that there would need to be a set of APIs from the DE, imho. I think that GNOME is going into this direction with their AI center, i whish KDE would start looking into that too
you know, i'd like to interact with a model directly in thunderbird so i can reply to LLM bullshit with LLM bullshit more more quickly
I've been using Aerion (https://aerion.3df.io) for a few months now, it's another mail+calendar+contacts app for Linux that is largely vibecoded. I haven't tested Penguin thoroughly yet, but I think Aerion is more feature rich, and has served me well over the months.
Why am I not using Thunderbird? Aerion's UI is much better, and most importantly snappier, especially on my older laptop.
Interesting that they call out Fastmail as a supported provider, but don't support JMAP. Seems like that would have been an easy stretch goal for Claude, and a good way to stand out amongst the competition.
On the other hand, we focus _only_ on JMAP. https://www.ariamailclient.com
I wouldn't say it's something that can be vibecoded in a month or two. Proper design and implementation take time if you want to handle accounts with hundreds of thousands of messages seamlessly.
I Feel so entitled and spoiled when I write this…
These days with the convenience of having LLMs at developer’s disposal, it feels unjustified to use electron like apps.. whenever I see “one app all/multi-platform”, I feel discouraged to give it a genuine try.
I tried a bunch of email clients and then crawled back to mail.app + Calendar.app on MacOS simply for how native it feels.
More importantly, which language(s) and UI toolkit are you using for your Linux app?
not to mention jmap is a breeze to use
The industry standard wording is to suffix the headline with "written in Rust".
Example:
"Penguin Mail – open-source email client for Linux with AI, written in Rust".
Its important, or people might miss the most interesting/important thing about your software.
This one should probably be "written in Rust by Claude".
3 years ago written in rust meant it was probably faster and smoother than other options. Now it means vibecoded
https://m.youtube.com/watch?v=qmVnr7rsWrE&pp=0gcJCY8Co7VqN5t...
Does anyone remember Sparrow?
I like the idea of a native Linux email client. The UI looks pretty good too. It's really not hard to beat thunderbird.
Calling it version 1.0 is bullshit though. It's only about a month old: https://github.com/c9dev/penguin-mail/graphs/contributors?al...
A project with the complexity of an email client cannot be considered stable in a month. There are bugs that will only show up after a year. It has a calendar after all.
> The UI looks pretty good too. It's really not hard to beat thunderbird.
I don't understand this mentality. It's an email client. There's not a lot needed UI-wise to make a functioning interface. Display a list of emails, show the full message when you open it, etc. I don't see anything functionally different between this email app and basically every other one I've ever used.
You've either got insane standards or are using email way different than I.
Can’t speak for others but for me Thunderbird’s UI feels distracting and almost grating in some ways. It’s technically functional but doesn’t feel right.
I know things like that don’t bother some folks and that’s fine, but others like myself can’t turn the visual parts of their brains off and so some software feels like nails on a chalkboard to use.
> It’s technically functional but doesn’t feel right.
That might be the most accurate description of Thunderbird that I've come across. It's hard to pinpoint exactly what's wrong with Thunderbird, despite me using it every day. Somehow it feels like the emails aren't the important bit of the UI, but I also couldn't tell you what I'm looking at instead.
The search is also incredibly slow. I didn't notice just how slow and inaccurate it is until I built my mail client on notmuch.
Maybe if you could spend some time defining the "feels right" aspect, this would go away.
I often find myself in situations where someone wants to break the UI I like, because they're used to something else.
It seems that most of the time, "feels right" just means "it works like I'm used to in other software". So: GMail, Outlook, whatever. Which is sad.
I wonder how would you react to "TheBat!" or Pegasus Mail?
> There's not a lot needed UI-wise to make a functioning interface
I remember waiting for ~15 years until Thunderbird finally implemented rich content support in the e-mail list (needed for a proper 3-column layout, as otherwise the e-mail list, being a simple table, wouldn't fit on the widest screens available).
So, yeah... Not a lot, but Thunderbird has been missing the bar for over a decade.
>I don't understand this mentality. It's an email client. There's not a lot needed UI-wise to make a functioning interface.
I don't understand this mentality.
Human communication is one of the most complicated activities in existence. Most (all) clients are modelling it very poorly.
Thunderbird still doesn't have proper threading
From an engineering standpoint, I get it.
But from a normal user's standpoint? I don't care. If my email client needs more than a single thread, something's fucked.
I randomly bounce between Thunderbird and KMail and honestly I can't tell the difference except cosmetically. Both are snappy. Neither crash. Both send and receive emails just fine.
Threading as in grouping together all replies to an email. This is still somewhat buggy on Thunderbird in 2026.
I know parent is probably talking about a different type of threading, but Thunderbird's lack of process threading means the UI just gets stuck when it's syncing a large numbers of emails, which can happen if your subscribed to a lot of mailing lists, etc.
Clearly you have never tried to use Thunderbird...!
1. Being released on GitHub for a month does not mean it’s a month old.
2. Seems odd to assume instability by default without evidence.
Have you ever experienced stable-by-default?
If so, I'd love to hear a writeup!
Well, I'm trying it out and it seems stable.
It's missing some features that are dealbreakers for me, but I haven't found anything unstable yet.
So it's probably missing features which should make a good 1.0 mailer.
> Calling it version 1.0 is bullshit though.
There is a case to be made to always start at 1.0, especially libraries with public API’s. The reason is that if you use semantic versioning then below 1.0 you can’t properly declare a new release to be a feature or bugfix release (for example, 0.4 to 0.5. Is this bugfix or feature?). Above 1.0.0 this is easy (for example 1.1.0 vs 1.0.1).
In libraries, 0.x is for when you’re still figuring the API out. Basically you’re saying that no API is stable at this point, and you’re not doing semver yet, and anything can change.
Naturally, the users should treat those libraries as beta versions and only use them in production at their own risk.
> you can’t properly declare a new release to be a feature or bugfix
0.5.0 vs 0.4.1?
See also https://0ver.org/#notable-zerover-projects
Nice UI but untrusted email messages is something I'd rather open in [an uglier app with decades long reputation for being reasonably secure](https://www.thunderbird.net/en-US/)
Would not be a problem with me, download source, let the agent search for backdoors and security problems for some hours and it would be fine for me. I would need to do the same with thunderbird if I'd need to trust an UI frontend to my email.
Thunderbird won't suffer from prompt injection
Not sure if this is gaslighting, an absurdist joke, or we are going through a collective psychosis.
KindOfCoders, AI is a very useful tool, but nothing works like the way you're explaining. I am kind of finding it hard to articulate the problem, because the assertion is just a bit wild. So I will do my best.
Thunderbird has been in production use for over 20 years, many many smart people have put their time into it to make sure it works correctly, yes, it is a bit clunky, it has a lot of craft, but you're not going to replace that much labour with running an AI agent for a few hours; despite the AIs usefulness in finding bugs and vulnerabilities. Not yet at least.
You are correct; however, agents can and do find terrible bugs in some of the most popular software in the world made by some of the most competent engineers on the planet.
I assume what KingOfCoders suggested is to run an agent almost like an antivirus which I guess sounds somewhat interesting but you'd have to re-run it on every update and probably spend many more hours+tokens for such a system to work properly.
But perhaps once inference gets truly cheap repository maintainers should include AI security checks for new software as part of their pipeline? I think that would make sense. Especially for repos that are frequently pwned like npm!
“With AI” means they have a built-in assistant (which no one wants), rather than exposing the functionality through MCP.
In the readme it says that MCP servers are available, https://github.com/c9dev/penguin-mail/blob/main/README.md#th...
No, it does not say that. It says you can connect MCPs to the built-in assistant, rather than using your own assistant + mail client MCP.
Designing an MCP is just slightly easier than designing a good API (you're not pressed to make things right from the first try).
Unlike slapping an AI agent on top while making zero promises about its capabilities.
All other email clients are crap, but instead of helping them, more half-baked attempts to reinvent the bicycle.
How so? Any specific critique? Ive been using Thunderbird tot décades and while e-mail seems to have some inherent clunkiness I find it overall quite fine.
How do you set it up for IMAP NOTIFY? Which GUI for ManageSieve do you use?
mu4e, mail.app, thunderbird. All pretty good IME.
How do you set them up for IMAP NOTIFY? Which GUI for ManageSieve do you use?
I think this looks good, but I'm not on a Mac anymore.
For people that want a terminal client with tabs, integrated html viewer(including images), markdown compose, MCP, with unified inbox, I built notmutt.
https://fishman.github.io/notmutt/
I’m confused, is this a joke about how real Linux users only use console, or did you miss tbe part where it says it’s for Linux?
He’s saying good looking GUIs are for plebian Mac users, you should use his vibeslopped TUI instead, like a real Linux user
“Email client” is an interesting puzzle. Email involves multiple activities:
Viewing: I have a message or messages stored on disk in files or mboxes or a maildir. How do I render them? Example use case is evidence in a law suit or, more likely, emails I’ve put in a shared drive to help me and my partner argue with our landlord. Yes, I could just have a Gmail folder for these but I don’t want to have one foot in a file-filesystem and one in a mail-filesystem. I want “lease_2024.pdf” and “20240401_deposit_confirmation_002.eml” to both be files. But then of course we have…
Filesystem interface: Emails have important attachments with MIME file names. Or perhaps a more keen observation is that some named files have emails as attachments! I’d like to symlink into a virtual FS that keys files off a UUID / Message-ID, and be able to mount that virtual FS from any mail store I have available (Gmail API, JMAP, IMAP, .tar archive, mbox, maildir, or .eml files.) Most conversations I have with people don’t involve files, and most files don’t involve conversations, but when they do I want them to feel natural and not some kind of special category of filey-messagey thing that doesn’t play well with plain files or, less commonly, plain messages.
Composition and sending: Creating a new message involves making a new file and then using that file to drive a sending API. I want to compose a beautiful email using a tool that is good at that job. Maybe I’m using vim to draft it in markdown with macros and tools to machine build certain complex parts? I also need some mime aware tooling — sometimes — to help prepare a bunch of attachments that may or may not have names related to their names on disk! I want all of this to be scriptable, on the occasions when it’s complex (30 part message for a client’s loan application!) and easy when it’s ad hoc (to: mom, see cat.jpg).
Search: email is notoriously the extension of and replacement for my long term memory. I have gathered message history from many sources and accounts over the years. I need to manage these into packed archives for the old ones, editable archives for the current ones, promote the former to the latter, and arrange new messages into the system. This feels like another separate tool.
I’m sure there are other potential tools lurking here as well. Separate rendering, composition, sending, receiving, searching, and archiving, feels like the goal. Most new email apps I see try to implement all of the above in one new package when, really, it would be great to see innovation on each branch of email as separate tools.
My PDF viewer doesn’t need to innovate on file browsing! My SVG drawing tool doesn’t need to innovate on mounting remote filesystems via SSH!
Tested on fastmail, here’s few notes from 10min impression:
- can’t choose a sender email?! This is major dealbreaker, if you have many aliases or domains.
- no favico for emails, this is also a deal breaker for me, as I used it to quickly skim and find or check new emails.
- folders orders is messed up, trying to arrange them made it worse
- trying to color a folder and it says “server doesn’t support that”
Also the UI needs more polishing overall.
> - can’t choose a sender email?! This is major dealbreaker, if you have many aliases or domains.
Ugh, this drives me nuts. It's the same reason why I stopped using my Supernote Nomad for email.
I use iCloud+ email with a custom domain. With that arrangement you authenticate to Apple's servers with your iCloud "primary address", like someone.somename@icloud.com. Then you can send and receive email with your custom domain email address, which is the one you're actually using. But not with Supernote, or apparently Penguin Mail. My Nomad will only send mail with my @icloud.com address, even though that's only used for authentication.
So yeah, I just stopped using it altogether. That's not my address. It's not how people reach me. It's literally just a username.
You can authenticate with your alias as well
Thank you for testing on Fastmail for us.
I'll stick with the web UI in that case. Are you using any other clients on Mac or Linux?
I am using their official linux client, which is basically the web one in app form with notifications.
wow that looks super polished!
I was wondering what “with AI” means and the page doesn’t really answer that thought the README does. It drafts replies.
Also it is made by pivotd ? What is it ? What is the end game ?
Genuinely curious as I am in the mail space with klar.im
This is interesting, I am doing an experimental effort with similar goals. What is your approach to testing? The bug surface of something like this is quite large. Also, do you have a plan for sandboxing?
Sandboxing should be handled externally by something like Flatpak
Surprised to see a new (large) GTK app, especially in Rust.
Can this be compiled for other operating systems than Linux?
Why? Most new gtk apps are written in rust now.
It's vibecoded - new, large codebases are par for the course.
(This used to come at the cost of code quality, but increasingly does not.)
I was 100% interested until crossing "...with AI".
I would love an email client that automatically sorts my mails via a local LLM.
As long as it tries to be the "I" in "AI" and do something useful like automatically detect conversations with my accountant, per project and not per sender mails including somehow labeling mails referring to multiple projects with multiple labels etc.
Not that I think this mail client can do that...
Second that. I can’t really think of a better application for LLMs than this. In fact, this might be the only good application.
Drafting emails with LLMs is... ugh.
And assistant chat? Really?
Okay, two things might be great: translations (preferably local, e.g. using Project Bergamot) and message summary (in case somebody does send an unedited blob of LLM slop to me and social dynamics don’t let me send it to spam).
If you took the time to read, it is an optional feature and uses local models.
Then my 'amail' client seems to fit the bill.. (OSS, rust, I use it on Linux and haven't tested elsewhere) ..not really suggesting it – it's pretty barebones and just whatever I've needed, but you're welcome to it obviously! Especially if you have issues/PRs to spur me on to features I've wanted too...
Same here. While I'd rather not have an embedded AI chat window, is there a way to disable it?
The homepage says “off by default”
Is it perhaps because of the application speed in Rust?
Got this building on FreeBSD for those interested:
1. Install the GTK stack:
2. FreeBSD ships WebKitGTK 2.46.6 and the project turns on the v2_50 feature of the webkit6 crate, so switch that to v2_46: After that, cargo build completes against 2.46.Penguin Mail is a mail and calendar app for Linux, written in Rust with GTK 4 and libadwaita. It's designed to be a lightweight, privacy-first alternative to Thunderbird or GNOME's built-in email.
What it does: • Aggregates Gmail, Microsoft (Outlook/Hotmail/Microsoft 365), Fastmail, iCloud, Yahoo, and any IMAP/POP3 account into one inbox • Calendar support for every account (Google Calendar, CalDAV) • Contacts with CardDAV integration • OpenPGP and S/MIME signing/encryption — keys never leave your machine • Gmail-style categories bar (Primary, Updates, Promotions, Social) • Send Later, Undo Send, templates, Markdown compose • A built-in AI assistant that can summarize threads, draft replies, manage rules, and search calendars. It runs on a local model (Ollama/LM Studio), an Anthropic API key, or Claude Code — you pick per job • Remote content blocked twice (WebKit filter + CSP), JS off by default • Runs in the tray at ~55 MB
It's GPL-3.0 and available as .deb, .rpm, Arch package, snap, tarball, and Flatpak. Not on Flathub yet — that's next.
We built it because existing Linux email clients felt bloated or didn't respect privacy. The AI assistant is a nice-to-have but not required — the app works great without it.
Feedback welcome: install experience, missing features, anything confusing. Happy to answer questions about the Rust/GTK4 architecture too.
Yet another GTK4/libadwaita, no thank you. I mean, if you're going to use Rust and vibe code thing, why not use a more sensible, Rust-native UI like egui?
Is there anything like this but in TUI?
Not visually similar to the app discussed here, but some TUI options:
https://terminaltrove.com/herald/
https://terminaltrove.com/matcha/
And this is the one I use with Gmail, Outlook, Yandex:
https://terminaltrove.com/aerc/
https://terminalemail.com/
Maybe I'm not the target here but an embedded chat ai window is a huge turn off to me. Having a CLI I can give my own agent to access/query the data, if I choose to do that, would have been preferred but if the target was my mother, then a chat window is probably a better option.