A tiny pixel crab that lives on your Mac, and quietly gets useful.

I've built Clawd, a free, open-source desktop companion for macOS. He's a little pixel crab who wanders along the bottom of your screen, says silly things, and can be dragged up into the menu bar to live there.

He started as a fun idea, the mascot from Claude Code living on my own screen, and kept growing into something with real jobs to do.

Clawd is built by me, developed with TIZZLE and owned by Crooli. He's an independent fan project inspired by the Claude Code mascot, and he isn't affiliated with Anthropic.

The idea

Most desktop pets are just that: a sprite that walks around and does nothing else.

I wanted one that earns its place on the screen. It should be small, charming and out of the way, but also able to tell you when something needs your attention and take small jobs off your hands.

So Clawd is drawn from nothing but filled rectangles. The same drawing code makes his sprite, his menu bar icon, his timer face and the animations on his website, and everything else is built around that.

He lives in your menu bar

Drag Clawd up to the menu bar and let go, and he shrinks into a status icon that blinks, walks in place and pops up short phrases. Drag him back out to carry him down to your desktop, or click him for his menu. Reminders and alerts show up as a small bubble under his icon, so he stays put.

He's notch-aware too. macOS quietly drops menu bar icons when there isn't room for them, so Clawd checks that his icon is really visible and, if not, hops back out and tells you to make some space.

He turns into a timer

Start a focus session or a break and Clawd becomes a tiny alarm clock, with a pixel LCD, a draining progress bar and a little bell on top. The same face sits in your menu bar with the countdown beside it, and when time's up a little bubble appears under his icon.

Drop things on him

Drag any file over Clawd and a tray of targets appears. You can hold it on a shelf, convert images to PNG or JPG, shrink them, zip files together, copy a path or send something to the Trash.

The shelf never moves your files. It just remembers where they are, so you can drag them back out wherever you like.

He tidies folders

Drop Clawd onto a Finder window and click Organise. He tosses every loose file into the right bin (images, documents, audio, video, installers, screenshots and so on) while the counts climb.

I was careful with this one, because moving people's files around is not something to get wrong:

• He only touches loose files at the top level of the folder you pick
• Subfolders, apps, hidden files and half-downloaded files are left alone
• He refuses code projects, your home folder and system folders
• Nothing is ever deleted, and Undo last organise puts every file back

A Claude Code companion

Clawd watches for a busy claude process, so he can tell you when a long run has finished.

There's also an optional hook that you install with one click (it backs up your settings first). With it, he waves and bounces when Claude Code is waiting for your permission or input, even while he's tucked away in the menu bar.

Small tools for developers

• A GitHub dashboard hangs off Clawd with your recent repos, CI status, open PRs and the reviews waiting for you. It uses the gh CLI, so there's no token to paste.
• A Git buddy keeps an eye on your local repos and nudges you about uncommitted work.
• A port watcher tells you when your dev server goes down.

Little shortcuts

There's a handful of everyday tools behind keyboard shortcuts:

• Screenshot to text: drag a box around anything and the text lands on your clipboard (it runs on-device)
• Colour picker, a calculator that converts units and currency, a password generator and quick notes
• Drop a YouTube or Spotify link on him to save it as an MP3 or MP4, using the open-source yt-dlp (only grab things you have the right to keep)

And he's a toy

Clawd dresses for the weather, wears a nightcap after dark, climbs onto your windows, and every so often does something strange, like building a tiny fort on top of a window or peeking at you from the edge of the screen. Every sound he makes is a soft 8-bit blip that you can mute.

He has a few games too: Token Catch, Flappy Clawd, a physics ball you can throw around your screen plus a basketball game where farther shots score more.

Getting him

Clawd is free and open source under the MIT licence, and works on macOS 14 or later.

The easiest way to avoid macOS's security warning is to build him yourself. It takes four commands and no Xcode project:

git clone https://github.com/xandertaylorx/clawd-buddy.git
cd clawd-buddy
./build.sh
open build/Clawd.app

If you'd rather just download him, there's a ready-to-run build on the releases page. It isn't notarised yet, so macOS will ask you to right-click and choose Open the first time. After that he can update himself from his own menu.

Meet Clawd

The code is on GitHub. If you try him, I'd love to hear what works, what doesn't and what you'd like him to do next.

Clawd is built by me, developed with TIZZLE and owned by Crooli. You can find more of my work at xandertaylor.org.

Learning to code or do maths shouldn't start with an account, a paywall or a wall of jargon.

I'm working on a new site called Stride, a gentle, account-free learning space for Maths, Python and Java.

It's still in progress, but it's far enough along that I want to share what it is and why I'm building it.

The problem

A lot of learning platforms ask for an email address before you've learned anything. Others throw you into content that's either too easy to be interesting or too hard to be useful, and when you get stuck the answer is usually to repeat the same thing until it sticks.

I wanted something calmer: you open it, you start, and it works out where you are.

How Stride works

Each subject begins with a low-pressure six-question starting check. There's no score to be embarrassed by. It's just used to build a personal route through the material, aimed at the same six foundation skills for everyone.

From there you work through authored topics:

• Java: 26 ordered topics across five phases, from first programs to architecture and collaboration
• Python: 17 topics covering lists, text, dictionaries, errors, files, testing, classes, recursion, algorithms and APIs
• Maths: 18 topics from negative numbers and ratio through algebra, graphs, geometry and statistics to growth and modelling

Each lesson separates the explanation, an interactive trace, a prediction, a modification, a connecting idea and an independent task, so you aren't just reading.

Evidence, not just ticks

I care about honest progress. Stride tracks three levels: introduced, practised, and used independently.

Getting something right after hints or retries doesn't count as independent. You need to apply it in two different situations on your first attempt, without help. Repeated mistakes never reset your place or trap you in endless remedial loops. They just lead to a fresh, different follow-up.

Meet Pip

Pip is the small companion who lives in each lesson. It gives hints, explains the actual compiler or runtime error you hit, points at the line that went wrong and suggests experiments to try.

Pip is deliberately deterministic in the lessons. It doesn't pretend to understand arbitrary code with AI. There is also a separate, minimal chat page where you can talk to Pip, limited to short replies so a small API allowance goes a long way.

A real playground

The Playground lets you run code in Python, Java, JavaScript, TypeScript, C, C++, C#, Go, Rust and more. Code only leaves your browser when you press Run, and the page tells you which external service it's going to before it does.

For Java there's also a portfolio workshop with a downloadable Maven and JUnit starter project with deliberately failing tests, so you practise the way real projects work.

No accounts, still on every device

Everything is stored locally in your browser. You can export a progress code and restore it elsewhere.

If you want your devices to stay in step, you can turn on sync. It creates a random code, encrypts your save in your browser, and stores only the ciphertext. There are no emails or passwords, and the server can't read your progress.

Building it

Stride is a build-less static site made of plain ES modules with no npm dependencies, deployed on Vercel. It has a test suite covering the placement logic, progress codes, sync merging, the playground and more.

What's next

There is still a lot to do. The advanced Java topics need more reviewed practice, and I want to keep improving the content quality before calling any of it finished. Stride is a pathway towards professional skills, not a certificate.

Try Stride

If you try it, I'd love to hear what works and what doesn't.

Stride is made by me. You can find more of my work at xandertaylor.org, and my studio, TIZZLE, builds websites and custom software.

Change your Java code, hit save, and see the results without restarting your application.

In collaboration with TIZZLE, I've developed Java HotSwap for Zed, an open-source extension that brings automatic Java HotSwap to what has quickly become my favourite code editor, Zed.

It's a relatively simple idea, but one that makes a massive difference when you're actually building software.

The problem

I've recently been developing my own Minecraft-inspired game using Java 25, Gradle and LWJGL.

One of the most frustrating parts of development was having to restart the entire game every time I wanted to test a small change. Something as simple as increasing the explosion radius of an arrow meant stopping the game, rebuilding it, launching it again and getting back to where I was.

Coming from Eclipse, I was used to being able to change Java code while an application was running. When I moved to Zed, I realised that the same automatic HotSwap workflow wasn't readily available.

I really didn't want to switch editors just for that.

So I decided to build a solution.

Introducing Java HotSwap for Zed

Java HotSwap allows you to modify compatible Java code while your application is running.

The workflow is simple:

1. Launch your Java application with HotSwap enabled.
2. Make a change to an existing method.
3. Save the file.
4. The extension recompiles and reloads the updated class automatically.

Your application stays running throughout.

For my Minecraft clone, I can now change the behaviour of exploding arrows, save the file and immediately test the new result without ever closing the game.

It's exactly the kind of development experience I wanted when I started using Zed.

Built for more than one project

I didn't want to create something that only worked with my own game.

The extension is designed to support Gradle, Maven and plain Java projects, including projects with multiple modules and builds nested inside larger workspaces.

It uses a Rust companion and a Java instrumentation agent to handle compilation and class reloading, while preserving existing build configurations and application settings.

There are some limitations. Standard Java HotSwap supports changes to existing method implementations, but structural changes such as adding fields or methods generally still require a restart.

The first release has been tested on Java 17, 21 and 25, with further compatibility testing planned.

Open source and what's next

The project is completely open source under the MIT licence.

This is the first working release, and there's still more I want to improve, particularly around compatibility, installation and making the experience as seamless as possible for other Zed users.

I'm also working towards making the extension available through Zed's extension marketplace. For now, it can be installed from the GitHub repository using the development extension workflow.

You can explore the source code, installation instructions and examples here:

Java HotSwap for Zed on GitHub

If you're a Java developer using Zed, I'd be interested to hear what you think. Contributions, bug reports and feedback are always welcome.

This project was developed by me in collaboration with TIZZLE, my digital studio focused on web design, development and custom software.

You can find more of my work at xandertaylor.org.

I wanted a cleaner split between the tech company and the studio.

I'm the founder of TIZZLE, the digital studio, and I'm introducing Crooli as a separate technology company.

That separation matters to me because TIZZLE and Crooli do different jobs.

Why I made the change

My work has grown beyond the digital studio I run through TIZZLE.

As I keep building software, online products, AI tools, and other technology projects, one brand started doing too much.

Crooli is the technology company.

TIZZLE is the digital studio.

Separating the two gives both a clearer purpose.

What TIZZLE does

TIZZLE is where I focus on client work:

• websites
• branding
• Shopify and e-commerce
• ongoing support
• search engine optimisation
• digital strategy and custom development

That means client work, case studies, service information, and enquiries now belong there.

What Crooli does

Crooli is focused on products, software, AI work, and new technology projects.

The work experience side of things stays under Crooli.

That part is still part of my story, and I want to keep it that way.

A new company, with a clear identity

This is not a change in ownership.

TIZZLE remains the studio I was already running, while Crooli gives my technology work its own home.

The studio's work, systems, and support stay the same. Crooli is a separate company for building technology.

What happens next

The technology company and its products are available at crooli.com.

If you want website design, development, or studio support from me, TIZZLE is the place to go.

The studio website, portfolio, service pages, and enquiries remain at tizzle.org.

Small plug: Crooli is my technology company, and TIZZLE is my digital studio. Explore the technology work at crooli.com or visit the studio at tizzle.org.

People feel your site before they understand your copy.

When someone lands on a page, they make immediate judgments:

• modern or outdated
• clear or confusing
• serious or cheap
• trustworthy or risky

Then they start reading.

Why that first impression matters

A great service can still lose if the website feels neglected.

If the design looks:

• messy
• slow
• broken
• outdated

people often assume the business quality matches.

Design is a trust signal before it is a visual decision.

Especially for smaller businesses

Big brands can survive weak sites because name recognition carries them.

Smaller brands usually do not have that margin. The website has to earn confidence quickly.

Clarity beats visual noise

You do not need flashy effects to look premium.

Focus on:

1. readable typography
2. clear offer explanation
3. real proof
4. obvious contact path
5. removing filler sections

Final point

Design will not rescue a bad service.

But good design can stop a good service from being ignored.

Trust starts before the first full sentence is read.

Small plug: At TIZZLE, I focus on websites that feel trustworthy at first glance and stay convincing once people read deeper. More at tizzle.org.

AI increases output speed. It does not automatically improve judgement.

AI lowered the barrier to producing almost anything:

• websites
• copy
• images
• code
• plans

That is useful. But it also means average work is easier to mass-produce.

The core problem

AI can scale quality, but it can also scale mediocrity.

You can generate polished-looking output that still feels empty.

Common examples:

• dramatic visual style with unclear message
• copy that sounds smart but says little
• logos with no personality
• content with no point of view

What still matters

Taste and judgement.

You still need to decide:

• what to keep
• what to cut
• what is generic
• what is actually useful for the user

Best use of AI

Use AI as leverage, not replacement:

1. speed up repetitive work
2. generate options
3. unblock blank-page moments
4. keep final quality decisions human

AI raises the floor. Taste raises the ceiling.

That gap is where strong work will continue to live.

Small plug: TIZZLE uses modern tools, including AI where useful, but the goal is still simple: better judgement, better design, better results. More at tizzle.org.

Ideas look best before reality touches them.

In your head, everything is smooth:

• design works
• users understand it
• launch goes well
• growth follows

Then you build it, and friction appears.

The perfection trap

Perfect ideas are often just untested ideas.

They have not met:

• edge cases
• awkward UX moments
• technical limits
• market indifference
• execution fatigue

A lot of people keep ideas “perfect” by never building them.

What execution reveals

Execution makes ideas honest.

You learn:

• what is too complex
• what users do not care about
• what sounded smart but feels clunky
• what is worth improving

First versions are supposed to be rough

That is not failure. That is entry.

Once something is real, you can improve it. Before that, you are guessing.

A better standard

I value boring execution more than exciting ideation.

1. build a real version
2. test it with real use
3. improve what matters
4. cut what does not

You do not need a perfect idea. You need a real one that survives contact with reality.

Small plug: TIZZLE is where I turn ideas into websites, tools, and digital products that can actually be used. More at tizzle.org.

Small useful tools are often better bets than big untested visions.

People talk like every idea needs to become a giant startup. I do not buy that.

Why small tools matter

Small tools solve one clear problem well.

Examples:

• calculators
• speed tests
• internal dashboards
• tiny workflow utilities

Usefulness beats hype.

Small tools are a clean product test

They force real product discipline:

• can someone understand it quickly?
• is the path from problem to result obvious?
• is it fast and focused?
• did you cut the unnecessary parts?

With big projects, it is easy to hide behind roadmap talk. With small tools, there is nowhere to hide.

Compounding effect

One tool may look minor.

Ten useful tools under one brand shows:

• taste
• execution
• momentum

It also shows you can finish things.

Practical upside

Small tools are easier to:

• maintain
• redesign
• use as experiments
• kill if they are not working

A shipped small tool in the real world beats a perfect giant idea in your notes app.

Small plug: TIZZLE builds websites, small products, and practical digital tools. The aim is simple: make things that actually work. See more at tizzle.org.

A website can be technically fine and still lose trust in seconds.

Most dead websites are not visibly broken.

• pages load
• buttons work
• forms submit

But they still feel flat.

What “dead” usually looks like

Common signals:

• vague copy
• generic imagery
• same layout as everyone else
• no proof
• no point of view

People do not always explain this with design language. They just leave.

The real problem

A website is often your first serious impression.

If it feels neglected, people assume the business is too.

This hits small businesses hardest because they do not have global brand recognition to compensate.

Signs of life

You do not need over-design. You need relevance and clarity.

Add:

• clearer offer statements
• real photos or real work examples
• case studies or outcomes
• a direct path to contact
• copy that sounds like an actual person

Simple vs lazy

• Lazy: text + stock image + button
• Simple: the exact message and action someone needs right now

That difference is where trust and conversion usually improve.

“Is the site live?” is no longer enough. “Does the site make us worth choosing?” is the better question.

Small plug: At TIZZLE, I care about making websites feel clear, current, and real - not like another forgotten template. More at tizzle.org.

The internet is huge, but most people still live on rented land.

Most people are known through platform profiles. Useful, yes. Owned, no.

Rented land vs owned space

A profile on any platform can change overnight:

• the algorithm changes
• your reach drops
• your layout gets redesigned
• your account is limited or removed

A personal website gives you control over structure, tone, and direction.

Why I still care about personal sites

For me, this site is a mix of:

• portfolio
• notebook
• public record of what I am building

I do not want something that reads like a generic corporate bio. I want something that feels human, specific, and current.

What a good personal site should do

A good site gives people context quickly:

• what you build
• how you think
• what you care about
• where you are heading

That matters if you want clients, opportunities, or credibility.

Keep it simple, but intentional

You do not need complexity. You need clarity.

A practical setup:

1. a clear homepage
2. a few strong projects
3. writing that sounds like you
4. one obvious contact route

The goal is not to look bigger than you are. The goal is to be understood faster by the right people.

Small plug: I build websites, tools, and digital systems through TIZZLE. If you want something sharper than a basic template, go to tizzle.org.

gardensfarm.uk is now live.

This post is less about hype and more about documenting the launch clearly.

Launch focus

The goal is to ship something clean and understandable for visitors:

• clear structure
• straightforward messaging
• easy contact path
• mobile-friendly experience

Why this one mattered

This is an early project where delivery quality matters more than looking impressive.

Reliable, clear work usually beats flashy work.

That mindset shapes the way I approach the build and the handover.

What I took from it

Going live is not the end of the story.
It is the point where the site starts doing real work for the business.

This launch reinforces a principle I want to keep following:

1. keep scope practical
2. communicate clearly
3. launch with confidence
4. improve from real-world usage

Appreciative of the trust on this project, and glad to see it live.

Small plug: I build focused websites and online systems through TIZZLE with the same approach: clear scope, clean launch, and steady improvement. More at tizzle.org.

Some milestones are better treated with gratitude than noise.

Today I sign Gardens Farm as my first client.

This is not a moment for big claims or celebration posts about "making it."
It is a moment to take the responsibility seriously.

What it meant to me

At this stage, I am still developing my process in the real world, not just in practice projects.

Working with a real business means:

• listening properly before suggesting changes
• respecting existing brand voice and relationships
• delivering work that was clear, useful, and maintainable

What I learned early

The biggest lesson is simple:

Good client work starts with trust, not features.

Before anything goes live, people need to feel that:

• you understand their business
• you communicate clearly
• you do what you said you would do

This project is helping me build those habits early, and I plan to carry them into every delivery cycle from here.

Final thought

I see this as a quiet but important starting point.

Not because it is perfect, and not because it is huge.
Because it is real, and it sets a standard for how I want to work with clients from day one.

Small plug: If you want clear, respectful delivery with practical communication from day one, that is exactly how I run projects at tizzle.org.