DeepakNess - Notes

Hardening a VPS with Tailscale

2026-06-08T03:28:52Z

I recently hardened this Hetzner VPS after realizing that keeping admin panels open to the whole internet is just unnecessary risk. And honestly, @levelsio gave me the idea to use Tailscale from a few posts he made about this on X.

On this VPS, I still have a few WordPress sites hosted through a control panel, and I wanted to make the server safer without deleting websites, moving data, or breaking normal access.

The main idea was simple: websites stay public, admin access becomes private. Ports 80 and 443 stay open because visitors need them. DNS, public web traffic, nothing changes from the outside.

But the control panel on its admin port, the web server admin panel, and SSH don't need to be open to every random scanner on the internet. Those are management interfaces, not public website features.

Tailscale gives the VPS a private network address that only approved devices can reach. So I can open the control panel from my computer through the Tailscale IP, but someone scanning the public VPS IP cannot hit the login page anymore.

The important part is the order. Don't close public SSH first – that's how you lock yourself out.

The safe order I followed:

take a snapshot first
then install Tailscale on the VPS and add it to your account (disable key expiry)
install Tailscale on your local machine
test that the control panel, web server admin, and SSH all work through the Tailscale IP
disable anything unused like FTP
restrict the admin ports
and only after all of that, restrict public SSH

For Tailscale, I used --accept-dns=false because I didn't want it changing DNS behavior on the VPS. This server hosts websites and mail, so the change needed to be narrow: add private access, don't surprise anything else.

Before changing any firewall rules, test everything over the Tailscale IP first. Confirm SSH, the control panel, and the web server admin all work. Once that's solid, allow those ports from Tailscale's private IP ranges and remove the public access rules for them.

For SSH, I used a rollback safety net: a systemd timer that automatically reopens public SSH after 3 minutes if I got locked out. After removing public SSH, I immediately tested both Tailscale SSH (should work) and public SSH (should fail). Only after confirming Tailscale SSH works, cancel the rollback timer.

A few things to remember: keep cloud console/rescue access available as a backdoor, keep Tailscale running on your machine before trying SSH or admin panels, firewall rules aren't active until you reload, and always test websites after any firewall change.

This isn't the same as rebuilding a compromised server. It's a practical hardening pass. But for a VPS that needs to keep running websites, it's a big improvement: keep the public web public, and make the sensitive admin stuff reachable only from your own devices.

Using the Composer 2.5 model in Pi agent harness

2026-06-02T14:25:47Z

Cursor released the Composer 2.5 model a while ago, and it's very capable model while being super cheap. And since I also love the Pi coding agent, I wanted to use the Composer 2.5 model inside Pi and found the perfect solution for it.

It's called pi-cursor-sdk and is created by Mitch Fultz. It uses the official Cursor SDK so Cursor's agent loop and tools are still used when using the model via the Pi agent. And it's a great thing because Cursor has one of the best agent harnesses out there.

I have been using it for the last few days, and it works flawlessly.

Trying the new MiniMax M3 model via OpenCode

2026-06-02T06:11:23Z

MiniMax M3 was released a day ago, and currently it's available in OpenCode for free for the last 2-3 days (yes, from even before the release). I, finally, gave it a try and it's so good, at least at UI design.

For example, I created this website using just two prompts (will share below) and it looks stunning. I am damn sure that Codex could never do this, and it's just Opus 4.8 level design but a lot cheaper. I have also taken a full page screenshot that you see here.

And if you're curious, prompts were:

Prompt 1:

create a stunning personal website for a developer. use dummy info. don't use any framework.

Prompt 2:

make it in light mode and don't use serif fonts. and make the design and ui even better.

After seeing it work great at the UI design, I asked it to redesign the entire dashboard for a new project I am working on, and it did a great job. I approved the plan and left it overnight, where it should have worked for a few hours and then when I woke up in the morning, it was ready. It had several minor errors which Codex quickly fixed, and I loved the overall UI and UX it did. I will soon be merging the new branch to the main.

For all these things, I used it for free via OpenCode (for a limited time), but if you wanted to try MiniMax M3, they have a very generous plan on their official website. Their $20/mo plan can get you up to ~1.6 billion tokens when you use the new M3 model. MiniMax has also released a desktop app called MiniMax Code and is currently available for macOS as well as Windows. It looks very similar to the Codex app, though.

By the way, since MiniMax M3 is an open-weight model, you can also try the new model via any other inference provider like OpenRouter and others.

Pi agent lessons from Dillon's shared session

2026-05-30T23:49:27Z

I spent some time reading Dillon Mulroy's shared Pi session, and honestly, I learned more from the workflow than the docs.

The main thing I noticed is that he doesn't use Pi like a quick code generator. He uses it more like a supervised senior implementation partner. The agent is not just asked to "build this thing". It is given specs, constraints, examples, review notes, and a clear process.

A few things I want to copy in my own Pi workflow.

1. Write the spec first

Before asking Pi to implement a large feature, create a small but clear tech spec with goals, constraints, routes, data model, edge cases, and test plan.

2. Add working-style rules to AGENTS.md

My current AGENTS.md has tool preferences and writing style rules. I should also add how I want to work – something like "prefer 3–5 small focused prompts over one large prompt", or "never rewrite a file without showing me the diff first".

3. Do architecture review before coding

Instead of directly saying "rebuild this", first ask Pi to inspect the existing files and explain where the design is weak, where cohesion breaks, and what should be changed.

4. Ask for call graphs and seams

Before implementation, ask Pi to sketch the public API, call graph, injected seams, and production vs test adapters. This makes bad design visible before code is written.

5. Use skills for methodology

A simple prompt like "use TDD" is weaker than a full TDD skill. Dillon injects an inline <skill> block that defines the philosophy, anti-patterns, and step-by-step workflow. I should create or reuse small skills for TDD, review, UI polish, refactors, and audits.

6. Use vertical slices

One failing test, one small implementation, one passing check. Don't ask the model to build the whole dashboard or whole feature in one go.

7. Give hard constraints

Prompts should include rules like: no new package, no mock data, no broad rewrite, no nullable return, no touching unrelated files. This keeps the agent on a tighter leash.

8. Use markdown annotations for review

Instead of explaining everything again, paste notes like this and ask Pi to apply only those changes:

## Markdown Annotations
- Section 4.1: update status to "implemented"
- Line 42: rename `fetchUrl` to `resolveShortLink`
- Section 3: remove the assumption about Cloudflare D1 – it's now R2

9. Codify mistakes permanently

When Pi makes the same kind of mistake, don't just fix it once. Add the rule to AGENTS.md so future sessions inherit it.

10. Use reference repos

For framework-specific work, keep local clones or examples and ask Pi to read those first. Real examples are better than model memory.

11. Use the session tree properly

If a design exploration is risky, fork the session. If it goes wrong, abandon that branch instead of trying to repair a messy conversation.

12. End with a handoff

At the end of a longer task, ask Pi for what changed, tests run, files modified, open issues, and the next 5 todos.

This is probably the biggest takeaway for me:

Better agent results come less from one perfect prompt and more from a better engineering loop.

The loop should be something like:

review the existing design
→ sketch options
→ agree on the shape
→ write or update the spec
→ implement one vertical slice
→ run checks
→ review with annotations
→ save new rules in AGENTS.md
→ hand off clearly

I already use Pi with DeepSeek, web access, goal tracking, and vision proxy, and I wrote about that setup in my Pi coding agent setup post. That setup is working nicely, but Dillon's session made me realize that it is still mostly tool-focused.

The next improvement is process-focused – better specs, better review loops, better constraints, and better handoffs.

Cheaper coding models are excellent with good prompts

2026-05-29T12:17:10Z

Dan posted about cancelling Cursor after 30 minutes of using Composer 2.5, and while the wording was too harsh, I don't think the frustration itself is completely wrong.

I posted about this on X as well, but I think the main thing people miss is that cheaper or smaller models need to be prompted differently.

You can't prompt DeepSeek, Kimi, GLM, Composer, Qwen, or other similar models the same way you prompt Claude Code or Codex.

With Claude Code and Codex, especially with the latest models, you can throw a broad task at them and expect the model to explore the repository, understand patterns, implement the feature, run tests, and come back with something mostly correct. It will still make mistakes, obviously, but it can handle much more ambiguity.

But these cheaper models are not great at that kind of one-shot work.

If you ask them to "build this full dashboard", "rework the entire app", or "implement this complete feature end-to-end", the result is usually messy. They will often do too much, miss existing patterns, break small things, or create a UI that looks good in one part but doesn't fully fit the app.

But if you go feature-by-feature or page-by-page, they can be excellent.

For example, instead of saying:

Build the entire analytics dashboard with filters, charts, empty states, settings, export, and responsive UI.

I would rather ask:

create the empty dashboard layout
add only the filters section
create the first chart
add loading and empty states
make the table work
polish spacing and mobile layout
now review the whole page and fix rough edges

This is a different workflow. You need more patience. You need to be more involved. You need to keep the model on a tighter leash.

But the tradeoff is worth it in many cases because these models are cheap and fast enough that iterating with them still feels good.

A few days ago, I built a small internal feature for SharePDF using DeepSeek V4 Pro via Pi agent. And honestly, the UI was far better than what Codex usually gives me, and everything worked as expected as well. But I didn't ask it to rebuild the whole product in one prompt. I gave it a smaller scope, checked the output, and then continued from there.

This is why I still think models like Composer 2.5, DeepSeek, Kimi, GLM, and Qwen are very useful. In fact, I have already written that Cursor's Composer 2.5 is awesome and that Cursor's $20 plan is great value. That doesn't mean the model will behave like Claude Code or Codex in every situation.

Sometimes when you have the option, there is no point using the cheaper model. If I'm doing a complicated refactor, touching billing logic, debugging something subtle, or making a change across many files, I would still prefer the best model I have access to.

But if I'm designing a page, improving UI, adding a smaller feature, or doing something where I can review each step quickly, I don't mind using cheaper models at all. In some cases, I even prefer them. I still consider Qwen 3.7 or even 3.6 Plus models far better at design than GPT models.

So I think the correct conclusion is not "Composer is bad" or "DeepSeek is bad" or "cheap models are useless".

The correct conclusion is:

don't use a cheaper model like a SOTA model.

Use Claude Code or Codex when you want more autonomy. Use DeepSeek, Kimi, GLM, Composer, or Qwen when you are willing to drive more actively.

Both can be great. You just can't expect the same prompting style to work everywhere.

Cursor's $20 plan offers the best value for month

2026-05-27T17:37:42Z

I last used Cursor in January this year, because it later started seeming too costly. But after their Composer 2.5 model release, I tried Cursor again after 5 months and I loved it a lot. I have been using it for a little over a week now, and it's the best model for the price out there.

At the time of writing this post, I have used over 430 million tokens in total where almost 320 million tokens are only from Composer 2.5, as you see in the screenshot above.

I have written a post on X about things I like about Cursor and then things I don't.

By the way, I am only on the $20/mo plan for Cursor Pro, and then I have only finished 54% of the monthly limits as of now. I am surprised how good the limits are even on the $20 plan, and I must say, again, that Composer 2.5 is a great model.

Fixing Cursor app's too much resource usage

2026-05-26T05:25:19Z

Cursor app had become weirdly unusable on my Mac recently, and I couldn't figure out why. So I asked Codex to do a read-only audit of my computer and create a small HTML report.

Turns out Cursor itself was not exactly the problem.

Cursor was launching an old Telegram MCP/plugin process from my ~/.claude folder, and that orphaned bun server.ts process was eating a lot of CPU in the background.

The funny part is, I don't even use that Claude setup anymore.

So I deleted ~/.claude, cleared Trash, and still the process kept running. That was the interesting bit: deleting files does not stop an already running process.

Codex then traced the exact PID, stopped only that orphan process, verified it didn't come back, and updated the report with what happened.

Now Cursor is normal again.

The main lesson for me: agent config is not harmless. Old MCP/plugin setup can keep affecting other tools long after you forgot it exists.

Also, I still don't fully understand why Cursor was launching something from my old Claude folder in the first place. But this is exactly why stale agent config feels like technical debt now.

Cursor's Composer 2.5 is awesome

2026-05-22T20:16:44Z

I last used Cursor in December last year, and after hearing great things about the new Composer 2.5 model, I decided to give it another try after 5+ months. While my experience is mostly positive, I will be writing about things I liked as well as things I didn't like.

What I liked:

Composer 2.5 model is fast, even when using without the fast mode. Most of the things I work on, just take seconds and sometimes a few minutes, making it faster than Codex or Claude models I have used in the past.
The 2.5 model is cheap, cheaper than all its competitors, actually. I used ~100 million tokens in a day and that resulted in only ~7% usage of the monthly limit, as you see in the screenshot here. I have posted on X about it in detail.

For your information, I am subscribed to the Cursor Pro plan which costs $20/month. And they're currently running a 2x offer for the week, but I think it's still a great value for money even after the offer ends. When I last used Cursor, the $20 plan was basically nothing when using any decent model. But that's not the case anymore, Composer 2.5 is a great model and that makes Cursor a great subscription plan right now.

What I didn't like:

I'm using a MacBook Air M2 with 16 GB of RAM and the Cursor app is just consuming too much resources, so much that the laptop is constantly running hot. I have heavily used Codex or Claude Code app but never had this issue earlier, so it's clear that they need to optimize the app for performance.

This was so serious that I ditched the app, and currently using the Cursor CLI only. It's great, I am loving it, and have even created a custom /statusline for myself.

My advice for others:

If you're confused about whether you should subscribe to Cursor, I would strongly suggest giving it a try. The Composer 2.5 model is amazing and can do most of your work.

As per calculations from my usage, if you're using 50 million tokens per day by using the Composer 2.5 model, the $20 plan will last for ~24 days before you exhaust your monthly limit. And this is great for people working on one or two projects simultaneously.

And if you can get the $60/mo plan, it's even better.

My experience with Antigravity 2.0 (it's bad!)

2026-05-20T02:35:30Z

Google I/O 2026 happened and Google announced a bunch of interesting things in the event. One thing I was particularly interested in was the new Antigravity 2.0, as I loved using the tool and coded the first versions of SharePDF by using this only in December-January this year.

But the new IDE didn't live up to my expectations, and it's bad!

So... let's look at everything bad one-by-one for the new Antigravity app:

1. Couldn't log in for hours

As soon as they announced Antigravity in the event, it was available for download. I updated the existing Antigravity app, and it got logged out when the new app opened. I tried logging in multiple times, but I kept seeing auth errors.

I kept seeing the above screen no matter how many times I tried. After an hour or so, the authentication successfully worked but when I quit and reopened the app, it started having the same issues again.

But it finally worked when I used it after several hours.

2. Keep asking for approvals

I have set Auto Execution and Review Policy options to "Always Proceed" as you see in the screenshot below, but it keeps asking for approvals each time.

And this is just an example, I had to approve at least 20 times for a simple HTML page I asked it to create. I have given "Full Access" to it, but this is still happening.

At least this wasn't happening on the Antigravity 1.0, right?

3. Creates lots of unwanted HTML files

I asked it to create a personal website for me by taking the required data from my existing website, but to be precise this was the exact prompt:

Create a clean and minimal personal website for me.
Take whatever info you need from my existing website: deepakness.com

And while it created the website using HTML, CSS, and JavaScript, it also created a bunch of temporary HTML files in the folder. I confronted, then it said that those can be safely deleted.

4. No plan mode anymore

I don't see the plan mode anywhere in the IDE, the dropdown we had earlier for planning/fast modes, no longer appears to be there.

5. Creates SPA all the time

When you don't mention the tech-stack, it always creates websites as SPA (single page applications). For the personal website sample project I created earlier:

All views (#home, #projects, #about, #newsletter) live in a single index.html and are toggled via JavaScript (navigateTo() in app.js) without full page reloads. It uses hash-based routing (#home, #projects, etc.) and content is dynamically shown/hidden client-side.

I tried the same prompt (from #3 above) at least 5 times, it created SPAs every time.

6. Has a bad design taste

I thought that the new model will have a great design capabilities, but it does not! It's still all purple, as you see in screenshots here.

I blame the Antigravity harness here, because designs are slightly better when I used the same prompt in the Google AI Studio.

7. Antigravity CLI also has bugs

To be honest, the Antigravity CLI is far better than the previous Gemini CLI and it's better than the Antigravity app as well. But it also has some bugs:

the prompt input field sometimes disappear
the scroll doesn't work at all sometimes
keeps asking for approvals, even after setting everything as "Always Proceed"

8. Other people sharing their experiences

Here are some posts by other people who have shared about Antigravity, that I found while browsing the internet:

Robin created a quick video on X sharing how bad his experience was.
The new 3.5 Flash model is very costly as well, as shared by someone on X.

Now, here are some things I liked:

I liked the multi-folder options for projects, as I can give access to multiple folders without giving access to the entire computer.
Usage limits for the new models are slightly higher than earlier on the AI Pro plan, and I think they have also added 1,000 extra AI credits for everyone.

That's it.

I will keep updating this page as I discover more things I like or dislike.

AI Search Optimization Skill as per Google's guidelines

2026-05-16T18:55:36Z

Recently, Google published a detailed guide on optimizing your website for generative AI features on Google Search, including a lot of interesting information. I came across this from this post on X from Gagan Ghotra, and created a powerful skill by consulting Gagan.

I published the skill in this public GitHub repo, and it can be accessed, installed, and used by anyone by following the provided instructions. And also created this quick video for beginners where I'm showing how to set up and use the skill (the same video on X).

Over the next few days, I will also be using the skills on different personal projects and will keep improving as per my experience with it. But since the GitHub repo is public, other people are also welcome to contribute.

And not to mention, the skill can be used with almost any AI agent/harness we have out there.

Freeing 100+ GB of storage on macOS using Codex

2026-05-16T13:02:31Z

I still use the MacBook Air M2 with 16 GB of RAM and only 256 GB of storage, and it was working perfectly fine for my use case until recently. I started getting the low storage warnings sometimes, and that annoyed me a lot. And I was about to factory reset my laptop... but then at the last moment, I decided to ask Codex about it, and glad that I did.

Codex helped me free more than 100 GB of storage by deleting unnecessary stuff that were there from some apps that I used and deleted earlier. For example, I once tried a local model via Hugging Face and LM Studio and they were together taking 35 GB of storage. Similarly, I tried a Docker alternative called Colima and it was taking another 19 GB, as you see in the screenshot here.

It also gave me suggestion to remove folders from the ~Library/Application Support/* from apps which were not used or not installed on my system anymore. From here, you can see that I try a lot of applications like browsers and so on.

Application Data on macOS" decoding="async" src="https://assets.deepakness.com/raw/codex-macos-cleaning/library-application-data.png">

The screenshots you see above, is from an HTML file that I asked Codex to create after analyzing my laptop in detail. And it created a complete checklist with all the information that I can do to optimize my laptop. It did a great job.

Now, the funny thing is I earlier used Mole and even subscribed to the CleanMyMac app and there also weren't able to free these unnecessarily occupied storage. I tried them a few times, and they only cleared a maximum of 5-10 GB storage.

Another funny thing is, I have used Linux for years in the past and I know about the file systems a bit. But I have become lazy ever since I started using macOS (I love my MacBook though) and didn't even bother to check those folders. But after Codex gave suggestions, I went and deleted unnecessary folders and files by myself, and cleared over 100 GB of storage.

Life's good now!

SharePDF now has 200 users

2026-05-12T17:37:28Z

I started working on SharePDF in January 2026, and it's been an amazing journey. It took ~4 months to reach 100 users, and now we're looking at 200 users in just 20 days after that.

Crazy, right? At least for me.

And in the last few weeks, I have also added enhanced existing stuff and have also added a few more features to the app. Some improvements done were:

Added a way to replace PDFs while keeping the same old share link
Introduced collections, a way to better manage lots of PDFs for power users
Improved analytics, so now users can see more data points, including per PDF analytics as well
Improved custom domain, billing, support, uploads, reliability, etc.

And still working to make everything even better for users.

Also, wrote a new blog post listing all the cool features SharePDF now has. I have tried to include everything here, and will probably keep this updated as well.

Apart from this, I am also launching SharePDF on Product Hunt tomorrow, i.e., May 13th at 12:30 IST. And I request your support for the launch, if you're reading this.

Thank you!

Run NVIDIA Kimi K2.6 and others in OpenCode

2026-05-10T07:41:09Z

I wanted to use Kimi K2.6 from NVIDIA inside OpenCode, and it was simpler than I first thought. This and a bunch of other open-source models are free to try in NVIDIA, by the way.

I only had to add the API key, model details, and the correct base URL in the OpenCode config.

I wanted to try Kimi K2.6 but you can try other models as well. For Kimi, NVIDIA's model name is:

moonshotai/kimi-k2.6

But inside OpenCode, I had use it like below

nvidia/moonshotai/kimi-k2.6

I edited this file on my macOS device:

~/.config/opencode/opencode.jsonc

And added this config:

{
  "$schema": "https://opencode.ai/config.json",
  "model": "nvidia/moonshotai/kimi-k2.6",
  "provider": {
    "nvidia": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "NVIDIA",
      "api": "https://integrate.api.nvidia.com/v1",
      "options": {
        "baseURL": "https://integrate.api.nvidia.com/v1",
        "apiKey": "YOUR_NVIDIA_API_KEY_HERE",
        "timeout": 600000,
        "chunkTimeout": 60000,
        "extraBody": {
          "chat_template_kwargs": {
            "thinking": true
          }
        }
      },
      "models": {
        "moonshotai/kimi-k2.6": {
          "name": "Kimi K2.6",
          "reasoning": true,
          "tool_call": true,
          "temperature": true,
          "attachment": true,
          "interleaved": {
            "field": "reasoning_content"
          },
          "modalities": {
            "input": ["text", "image", "video"],
            "output": ["text"]
          },
          "limit": {
            "context": 262144,
            "output": 65536
          },
          "options": {
            "chat_template_kwargs": {
              "thinking": true
            }
          }
        }
      }
    }
  }
}

The important bit is this URL:

https://integrate.api.nvidia.com/v1

NVIDIA's example uses the full /chat/completions URL, but OpenCode only needs the base URL. It adds /chat/completions by itself.

Works as expected, so that's it.

Laravel installer getting killed after update

2026-05-06T16:50:15Z

I ran into this after updating Laravel / Herd Lite using the php.new installer. The update itself looked successful, but when I tried creating a new Laravel project, the command was immediately killed by macOS.

Initially, it looked like a Laravel installer issue because I also saw a PHP parse error from inside the Laravel installer PHAR. But the actual issue was lower-level: the bundled PHP binary was getting blocked/killed when running from Herd Lite's installed bin directory.

The fix was to download fresh PHP, Composer, and Laravel installer binaries and place them in a clean local bin directory. Then I updated my shell PATH so this clean directory is used before the problematic Herd Lite bin path.

In short:

export PATH="$HOME/.local/herd-lite-bin:$PATH"

After reloading the terminal, php, composer, and laravel resolved to the working binaries, and laravel new started working again.

So if Laravel commands suddenly start saying zsh: killed after a Herd Lite / php.new update, check where your php and laravel binaries are coming from first.

From free tools to full SaaS

2026-05-06T01:10:16Z

I created a free tool a few months ago, and now it's getting 10-15 clicks every day from Google, as you see in the screenshot here. So... I researched a bit and realized that there is a solid SaaS opportunity on the site, I can have the main tool free but then can also build some paid features in the web app.

And not just this, take a look at another free tool I created before ChatGPT was launched, and it's still getting lots of traffic from Google and other referrers. Take a look at the screenshot here:

This one is getting 80-140 clicks a day consistently for years now, and I haven't done anything with it yet, but not anymore. I am working on it to bring some paid offerings in the app, currently, in the brainstorming phase. For research, I am analyzing other similar apps and also going through the data from Google Search Console, and it's been very helpful so far.

If you're wondering, I am not going to tell you what the app is about just yet because it's very simple, and the idea can be copied with just a prompt. But yes, I will eventually share more about these.

Honestly, "distribution" has always been important, but it has become even more important in the age of AI because "code" is not the moat anymore. In my case, I already have thousands of visitors coming to the tool and if I create the paid version, I will instantly have lots of potential customers.

Also, the concept of engineering as marketing is the most rewarding than ever in the current time, because creating free tools around your main product is simpler and faster than ever. If you want to experiment something with the lowest effort, just ask any LLM to give you ideas for free tools around you main product, create one, publish it in a subdomain, and forget it for a few weeks. If you see a positive traction, create the tool on the main domain, and do a redirect from the subdomain.

I think, this is the best use-case of AI for marketing.

Save a clipboard image to computer on macOS

2026-05-01T16:25:54Z

I take a lot of screenshots and most of the time keep them in my clipboard and then paste wherever needed, like on socials or in AI coding tools when working. But sometimes, I also need to save those images locally on my computer, and till now I have been doing one of these two things:

Either take the screenshot once again, or
Paste images somewhere like on Imgur, and then save by using "save image as" feature

Silly, but yes.

However, I recently found out a trick that works like a charm, at least on macOS. The process is:

Open the Preview app, you can do so by searching for Preview via Spotlight search
Press cmd + n and you should already see the image from your clipboard

Now, just save it using cmd + s or however you want to, as you see in the screenshot here.

It works as expected and even gives you image format you want to save your screenshots as.

TIL.

DeepSeek v4 Pro should be taken seriously

2026-04-30T17:01:42Z

For the last 2 days, I have been using the DeepSeek v4 Pro model via OpenCode Go and it has been a fantastic experience. I never expected the model to be this good at UI as well as logic.

For example, I am creating a custom blog publishing CMS powered entirely by Cloudflare (D1, R2, and Workers). Initially, I used Codex for this and Codex did ready the first working version, but the UI was very bad and cluttered and the user-experience wasn't great either. I didn't even like the CMS a bit.

But then I fired up OpenCode and gave everything to the DeepSeek v4 Pro model, and asked it to re-do and re-design everything by keeping user-experience in mind. I, then, went to the gym and the model had done its magic by the time I returned. It ran for ~35 mins and completed everything in one go, and the design was already looking much cleaner and just better.

I still had to give a few more prompts to refine some things, and it did everything perfectly. Just for your information, while working, I kept the thinking effort as default, as you see in the screenshot.

From the next month, I think I am going to save decent amount of money on AI tools subscriptions. Currently, I am subscribed to Codex $100 plan but now I think Codex $20 with OpenCode $10 (+ $10-20 for extra usage) should be enough for my use case.

Lastly, I think the DeepSeek v4 models should be taken more seriously by everyone. Their API pricing is also currently discounted by 75%, and there is absolutely no reason to not use this model.

Introducing EternalQR – a dynamic QR code tool

2026-04-28T03:24:11Z

Introducing EternalQR – a small web-app to create dynamic QR codes, update their links later, track scans, customize them, and download them as PNG or SVG files.

I got the idea for this from something that happened to my brother.

He had created a video on YouTube and added a QR code to it using some free QR code tool he found online. Initially, the QR code was working correctly. But later, when the video started getting lots of views, the QR code stopped working.

And because the QR code was already inside the video, there was no way to update or replace it.

That made me think that QR codes are a bit different from normal links. A link on a website can be changed anytime. But a QR code often ends up in places that are hard to edit later – YouTube videos, restaurant menus, business cards, flyers, packaging, event signs, invoices, stickers, and so on.

So I created EternalQR around one simple idea: if a QR code is already created and used somewhere, normal plan changes should not turn it into a dead link.

Some things it supports right now:

dynamic QR codes with editable destinations
scan analytics with country, device, browser, referrer, and time
QR codes for websites, contact cards, Wi-Fi, SMS, email, phone, and plain text
logo and color customization
PNG and SVG downloads
pause, expiry dates, tags, and basic dashboard management
existing QR codes keep redirecting through normal plan changes

The app is built on Cloudflare Workers, D1, and KV, so redirects are very cheap to serve. Paid plans are mainly for things like more QR codes, longer analytics history, customization, and other features.

The pricing is also kept simple:

Free – 5 QR codes
Starter – $5/mo for 50 QR codes
Pro – $19/mo for unlimited QR codes
Lifetime – $89 once for Starter features

Not to mention, but all these plans come with unlimited scans.

Here's how an already created QR code page looks, with scan analytics and more.

It's not a huge ambitious product, but just a practical tool for a problem I saw happen in real life.

Compact Tabs are back in the Safari browser

2026-04-26T20:42:27Z

I love using the Safari web browser on my MacBook, especially when I'm traveling because it consumes less battery on the laptop. But with the macOS Tahoe 26 update, the Compact tabs feature was removed from the browser, and I was sad about it.

And I'm just learning that with the new macOS Tahoe 26.4.1 update, the compact tabs feature is back in the Safari browser, as you see in the screenshot here.

It's good already, but I think it behaves a bit differently than it did in the macOS Sequoia or Sonoma, when it was still available. But it's still good, and I'm using it for a day and love it.

Thank you, Apple.

How to best spend $50 on AI coding tools

2026-04-24T16:24:36Z

I have been spending over $100 every month on Claude Max and Codex Pro plans as I use the latest models from these providers for coding, but now I don't think that's the best way to do it anymore.

The current best option I think is, instead of getting higher plan for any one AI tool, get smaller plans for multiple providers. And the best combo I can think of is the following.

Claude Pro – $20/mo
Codex Plus – $20/mo
OpenCode Go – $10/mo

Getting all these will only cost $50, but this is more than enough for my requirements. Here I will save money and also have access to top models from multiple providers. The different models can be used for different tasks that they're actually good at, for example:

for all design and UI related work – use the latest Claude models
for complex logic and programming – use the latest Codex models
for smaller tasks and everything else – use Kimi, GLM, or Qwen models via OpenCode

And not just me but several other people are also starting to realize this.

Currently, I am subscribed to Codex $100 plan that ends next month, and also to OpenCode Go plan. But I'm going to downgrade Codex to $20 plan and then also subscribe to Claude $20 plan and keep my OpenCode Go for $10.

That should be more than enough for three full-time projects I am currently working on.

My experience with the new GPT-5.5 model

2026-04-24T08:41:29Z

Yesterday, OpenAI launched the newest model GPT-5.5 which I was too excited about for the last few days. And here are my experiences so far with using the new model inside the Codex app.

GPT-5.5 is far better in frontend design than other OpenAI models like GPT-5.4 or 5.3-Codex. But it's still not at the level of Opus 4.7 of 4.6 when it comes to UI design.
Even though it was claimed by Sam Altman that the new model uses fewer tokens per task than previous models, it's not the case when I use it. In my case, it used way more tokens than GPT-5.4 because earlier I could never finish the 5-hour limit on the $100 plan, but today I did, as you see below.

The new model, GPT-5.5, definitely feels slightly faster than the previous, GPT-5.4, model in the standard mode. I think, both are equally faster in the speed/fast mode.
I didn't notice any significant improvement in the programming though with the GPT-5.5 model, it feels similarly capable as GPT-5.4 model. Or, maybe I haven't tested this enough yet.

Also, the new Codex app feels snappier and the UI and UX is slightly improved now.

Apart from this, I can still not rely completely on the Codex subscription itself. I can use OpenCode with models like GLM-5.1 and Kimi-K2.6 for frontend design and then GPT-5.5 for everything else. The OpenCode Go plan is very generous at $10/mo and has crazy limits as well. So if I had to reduce my monthly spending, I would go with OpenCode Go ($10/mo) + ChatGPT Plus ($20/mo) and maybe $20-30 extra on OpenCode, and that should cover all my requirements.

I'm still playing with the GPT-5.5 model so if I find something new worth sharing, I will keep this post updated.

Update: Apr 25, 2026

I posted about limits draining too fast on X and many people recommended that I should rarely use Extra High effort for my tasks, and suggested using either Medium or High for most tasks. So... I gave the "High" effort a try, and it's good, I am satisfied with it so far.

SharePDF reaches 100 users

2026-04-23T06:43:06Z

The app to share PDFs, SharePDF.app, that I created early this year has now reached the milestone of 100 users and I'm so happy for it. It took around 4 months to reach this milestone and I don't know what, but it makes me super happy.

I have been documenting most of the things on this page and reading previous logs takes me through the journey of things I did in the past few months – how I finalized which tech-stack to use, how I decided different features the app has, and more.

SharePDF is running stable and I don't need to do much on it now. But I still have 1-2 features or rather enhancements planned for near future. Also, I will be working on improving the landing pages as well, and maybe even create a few new ones targeting keywords that get searched a lot.

Codex is smarter than Claude Code

2026-04-21T07:07:11Z

After seriously using Codex for a few days, it definitely seems smarter than Claude Code. But Codex is just awful at design. I have tried to make good designs using Codex multiple times, and it just doesn't work. In fact, I tried creating a few new pages for my Vemgram project that I am mainly working on these days, and even though the site has a design system set up, Codex is still unable to create designs that matches the aesthetics of the rest of the website.

I am very much disappointed for that.

For this reason, nowadays, I have started using OpenCode whenever I need to design something. I love the newest launched models like Kimi K2.5, Qwen 3.6 Plus, and GLM-5.1 models inside OpenCode. These open-source models are almost as good as Opus 4.6/4.7 when it comes to design.

Apart from this, I am hearing rumors that OpenAI is about to launch GPT-5.5 and also that it's great at design. If that's the case, I won't regret subscribing to Codex Pro at all.

After Claude, now Codex can't keep up with the demand

2026-04-20T15:18:22Z

It's been almost an hour since OpenAI's Codex is completely down, as you see in the screenshot below. I recently moved away from Claude for this reason, and now I am facing the same issues here as well.

Codex still hasn't become as bad as Claude was a week ago and before that, but it seems like it's getting there. And it's surprising to me that if these AI companies can't keep up with the rising demands, why don't they just pause new signups temporarily? I mean, it will be better for everyone, no? – for users and then for their own brand image as well.

This is not at all fair for users. They're paying the premium to just experience degraded performance all the time. For example, just take a look at Claude's status chart for the last 90 days and you'll see lots of yellows and reds.

I, myself, had been paying at least $100 each month to Claude and now to Codex, but getting the same degraded performance. Honestly, I don't blame OpenAI or Anthropic for that they can't keep up with the demand, but I blame them because why are they still allowing signups and ruining the experience for everyone? They should just pause signups for some time till they get more compute, and if required pause again for some time till they get more compute, and so on.

I also run a SaaS, and I will feel very bad if I keep charging a fee to my customers and if my app is consistently down. I will even refund them and pause new signups till I get things sorted.

That's it.

Sorry for the rant, really.

I like Claude Code CLI's no flicker mode

2026-04-14T12:31:19Z

I just started using Claude Code's new no-flicker or full-screen mode for the CLI and I like it better than earlier. When enabled, it just sends the text input field at the bottom of the screen and makes it sticky. Just like how you see in the screenshot here:

I like it because it doesn't disturb me when Claude is generating a large output, then I can keep reading from the top while it outputs at the bottom, without disturbing my flow.

Mistakes were made at Google Search Console

2026-04-14T12:10:00Z

Today, I suddenly started receiving emails like below from Google Search Console for all my sites already added. These emails are normally sent when a new site is added to the GSC for tracking the search performance, but mistakes were made, for sure.

I have at least 8-10 sites added to this Gmail account and I received emails for all those in a 3-5 hour span. I also posted about this on X and several people confirmed that they have also received the same emails.

And if you're curious about the contents of these emails, see the screenshot above. It's the same that Google sends when you add a new site to GSC.

I hope they fix it soon and do not annoy me, again.

Update: Apr 15, 2026

It was just a "glitch" as everyone suspected, as confirmed by John Muller from Google. Learned about this from this post on X by Gagan Ghotra.

Introducing SharePDF app

2026-04-14T04:29:42Z

Introducing SharePDF – a web-app to host PDFs, share them as URLs, track their view analytics, and do much more. It's live and working, and already has more than 80 users at the time of writing this post.

I had this idea to create an app that makes sharing PDFs easier for at least 2 years, but I never worked on it. However, somehow I started working on it from January this year, i.e. 2026. I haven't done any kind of marketing as of now, and still regularly getting signups from sources I wouldn't even have guessed – Instagram and ChatGPT. Yes, someone must have created some content that people are discovering SharePDF from.

Today, I have also posted on X for the first time about the app. Apart from this, I have also been sharing build logs on this page on my website – it just helps me keep myself in the loop.

Currently, I am not working on new features for the app but mainly working on improving the landing page, copy, and user experience.

By the way, I also launched Vemgram, a platform to connect Indian manufacturers with buyers, a few days ago. So, now, I have two full time projects that I am working on these days.

Better utilize Claude Code limits

2026-04-11T04:01:14Z

Came across this post on X that shares a hack, a way to maximum utilize your Claude Code usage limits by setting up scheduled tasks. It seems silly at first, but it's actually helpful.

As explained in the post, I set up a cron that sends "hi" every 5 hours starting from 8AM IST so that as soon as the limit resets, it immediately starts a new cycle so that you have clean time blocks.

30 2,7,12,17 * * *

And the best thing is, you don't need any extra tool to use for this.

Meta's new Muse Spark model is here

2026-04-08T17:21:02Z

Meta just launched a new AI model called Muse Spark which is now available to use via meta.ai and the Meta app. It's not available via the API, yet.

As of now, it looks good on the evals as you see in the screenshot here. In categories, it's shown to be even better than Opus 4.6 and GPT-5.4.

I tried the model via meta.ai web interface with my Facebook login, and I must say that it has a better sense of design than GPT-5.4. For example, I just asked it to create a minimal personal website and this was the result.

I am still testing this, and will keep this post updated.

Also, I must say that I liked the meta.ai web interface and it feels fast and snappy. It even let me see the website in a new tab.

Milla Jovovich's mempalace is nice, but...

2026-04-07T16:13:12Z

Milla Jovovich launched an open-source AI memory system called mempalace with a bold claim saying "highest-scoring AI memory system ever benchmarked". She first announced about this on her Instagram and then it's picked by multiple folks from there.

And yes, that Milla Jovovich from movies like Resident Evil and The Fifth Element. She's an engineer and loves to code as well.

But then I came across this issue on GitHub that highlights multiple false claims, as you see in the screenshot.

As she claims in the video, it's still work in progress, and she might still be working on this, so there might be some gaps in the README vs the actual codebase. But the concept of the tool is still very good, and there is also lots of good information in the README file that you can explore.

Google AI Edge Gallery app is good

2026-04-07T15:40:25Z

Google has a new app for Android and also for iOS called AI Edge Gallery that lets people explore and use local on-device LLMs. Currently, it's featuring the newly launched Gemma 4 family of models.

This is written in their app description on app stores:

AI Edge Gallery is the premier destination for running the world's most powerful open-source Large Language Models (LLMs) on your mobile device. Experience high-performance Generative AI directly on your hardware—fully offline, private, and lightning-fast.

And to my surprise, the app itself is open-source. I love whatever is Google's game here.

I tried the app on my Android device, and even downloaded the smallest Gemma-4-E2B-it model and it worked fine. The app has multiple different options, as you see in the screenshots, and it looks stunning as well.

I am traveling next week via train and the network is not stable sometimes, so I can easily chat with these local models. I have tested for a bit, and they're good at non-coding stuff.

Blocking disposable email sign-ups in Laravel

2026-04-05T02:41:53Z

I noticed some users signing up on a project with throwaway emails from services like yopmail, guerrillamail, and tempmail. They'd sign up, poke around, and create multiple accounts to use the service for free.

I researched solutions for it and found propaganistas/laravel-disposable-email, a popular Laravel package (~1.35M downloads) that gives you an indisposable validation rule. It pulls from a community-maintained list of ~72k known disposable domains.

But I didn't want another package dependency for something this simple. So I grabbed their JSON file with 72,000+ disposable domains, and wrote a tiny custom validation rule:

class NotDisposableEmail implements ValidationRule
{
    public function validate(string $attribute, mixed $value, Closure $fail): void
    {
        $domain = strtolower(substr(strrchr($value, '@'), 1));

        if (in_array($domain, $this->disposableDomains(), true)) {
            $fail('Disposable email addresses are not allowed. Please use a permanent email address.');
        }
    }

    private function disposableDomains(): array
    {
        return once(function (): array {
            $path = resource_path('data/disposable-domains.json');

            if (! file_exists($path)) {
                return [];
            }

            return json_decode(file_get_contents($path), true) ?? [];
        });
    }
}

Then added it to the registration validation in CreateNewUser.php:

'email' => [
    'required',
    'string',
    'email',
    'max:255',
    new NotDisposableEmail,
    Rule::unique(User::class),
],

That's it. No package, no API calls, no latency. The JSON file loads once per request (using Laravel's once() helper), and if the file is somehow missing, it silently passes so nothing breaks. It shows this when someone tries signing up using a disposable email:

The only downside is that the list is static and new disposable services won't be caught until I update the JSON. But for now, it's good.

Update:

After many people pointed out on Thread and X, I have now removed the blocking for disposable domains and exploring better ways to filter out low-quality signups.

Backing up important GitHub repos

2026-04-04T12:08:24Z

Lately, GitHub hasn't been very stable and I also read a story of GitHub blocking someone's account on Reddit, so I do not have a lot of trust to keep my code only on one platform. And I used my small Raspberry Pi 4B 1 GB device to keep additional copies of my important repos.

Here's the bash script to mirror specific GitHub repos locally (with optional GitLab push). Supports org repos, wikis, LFS, allowlist/blocklist filtering, and dry-run mode:

First, this backup_github_repos.sh file:

#!/usr/bin/env bash
set -Eeuo pipefail

usage() {
  cat <<'EOF'
Backup all owned GitHub repositories to local mirror clones and optionally push them to GitLab.

Usage:
  ./backup_github_repos.sh [path/to/config.env]

Prerequisites:
  - git
  - gh
  - jq
  - gh auth login
  - gh auth setup-git

Optional:
  - git-lfs (if FETCH_LFS=1)
  - GitLab SSH key (if ENABLE_GITLAB_PUSH=1)

This script:
  - mirrors all repos you own on GitHub
  - can also mirror selected GitHub org repos
  - can limit backups to an exact allowlist of repos
  - can back up wiki repos when they exist
  - can push matching repos to GitLab over SSH

It does NOT back up GitHub issues, pull requests, discussions, or release assets.
EOF
}

log() {
  printf '[%s] %s\n' "$(date '+%F %T')" "$*"
}

warn() {
  printf '[%s] WARNING: %s\n' "$(date '+%F %T')" "$*" >&2
}

die() {
  printf '[%s] ERROR: %s\n' "$(date '+%F %T')" "$*" >&2
  exit 1
}

run() {
  if [[ "${DRY_RUN}" == "1" ]]; then
    printf 'DRY_RUN:'
    printf ' %q' "$@"
    printf '\n'
    return 0
  fi
  "$@"
}

trim() {
  local value="${1:-}"
  value="${value#"${value%%[![:space:]]*}"}"
  value="${value%"${value##*[![:space:]]}"}"
  printf '%s' "$value"
}

require_cmd() {
  command -v "$1" >/dev/null 2>&1 || die "Missing required command: $1"
}

csv_list_contains() {
  local needle="$1"
  local csv="$2"
  local item

  IFS=',' read -r -a items <<<"$csv"
  for raw_item in "${items[@]}"; do
    item="$(trim "$raw_item")"
    [[ -n "$item" ]] || continue
    [[ "$item" == "$needle" ]] && return 0
  done

  return 1
}

gh_repo_stream_owned() {
  gh api --paginate "/user/repos?affiliation=owner&per_page=100" \
    | jq -c '.[] | {
        name,
        full_name,
        private,
        archived,
        fork,
        has_wiki,
        clone_url,
        ssh_url,
        owner: .owner.login
      }'
}

gh_repo_stream_org() {
  local org="$1"
  gh api --paginate "/orgs/${org}/repos?type=all&per_page=100" \
    | jq -c '.[] | {
        name,
        full_name,
        private,
        archived,
        fork,
        has_wiki,
        clone_url,
        ssh_url,
        owner: .owner.login
      }'
}

write_repo_metadata() {
  local repo_json="$1"
  local owner="$2"
  local repo="$3"
  local metadata_dir="${BACKUP_ROOT}/metadata/${owner}"
  mkdir -p "$metadata_dir"
  jq '.' <<<"$repo_json" > "${metadata_dir}/${repo}.json"
}

sync_git_mirror() {
  local source_url="$1"
  local destination="$2"
  local label="$3"

  mkdir -p "$(dirname "$destination")"

  if [[ -d "$destination" ]]; then
    log "Updating mirror: ${label}"
    run git -C "$destination" remote set-url origin "$source_url"
    run git -C "$destination" remote update --prune
  else
    log "Creating mirror: ${label}"
    run git clone --mirror "$source_url" "$destination"
  fi
}

fetch_lfs_objects() {
  local destination="$1"
  local label="$2"

  if [[ "${FETCH_LFS}" != "1" ]]; then
    return 0
  fi

  if ! git lfs version >/dev/null 2>&1; then
    die "FETCH_LFS=1 but git-lfs is not installed"
  fi

  log "Fetching LFS objects: ${label}"
  if ! run git -C "$destination" lfs fetch --all origin; then
    warn "LFS fetch failed for ${label}. The Git mirror is still valid, but LFS content may be incomplete."
  fi
}

gitlab_remote_url() {
  local repo_name="$1"
  printf 'git@gitlab.com:%s/%s.git' "$GITLAB_NAMESPACE" "$repo_name"
}

gitlab_remote_exists() {
  local repo_name="$1"
  git ls-remote "$(gitlab_remote_url "$repo_name")" >/dev/null 2>&1
}

ensure_gitlab_repo() {
  local repo_name="$1"

  if gitlab_remote_exists "$repo_name"; then
    return 0
  fi

  if [[ "${GITLAB_CREATE_REPOS}" != "1" ]]; then
    die "GitLab repo ${GITLAB_NAMESPACE}/${repo_name} does not exist and GITLAB_CREATE_REPOS=0"
  fi

  log "GitLab repo ${GITLAB_NAMESPACE}/${repo_name} will be created on first push if your SSH key can create projects in that namespace"
}

push_git_mirror_to_gitlab() {
  local destination="$1"
  local repo_name="$2"
  local label="$3"
  local remote_url

  remote_url="$(gitlab_remote_url "$repo_name")"

  if git -C "$destination" remote get-url gitlab >/dev/null 2>&1; then
    run git -C "$destination" remote set-url gitlab "$remote_url"
  else
    run git -C "$destination" remote add gitlab "$remote_url"
  fi

  log "Pushing branches and tags to GitLab: ${label}"
  run git -C "$destination" push --prune gitlab \
    '+refs/heads/*:refs/heads/*' \
    '+refs/tags/*:refs/tags/*'

  if [[ "${FETCH_LFS}" == "1" ]]; then
    if ! run git -C "$destination" lfs push --all gitlab; then
      warn "LFS push failed for ${label}. Check GitLab LFS configuration if this repo uses LFS."
    fi
  fi
}

should_backup_repo() {
  local repo_json="$1"
  local archived
  local forked
  local full_name

  archived="$(jq -r '.archived' <<<"$repo_json")"
  forked="$(jq -r '.fork' <<<"$repo_json")"
  full_name="$(jq -r '.full_name' <<<"$repo_json")"

  if [[ -n "${REPO_ALLOWLIST}" ]] && ! csv_list_contains "$full_name" "$REPO_ALLOWLIST"; then
    return 1
  fi

  if [[ -n "${REPO_BLOCKLIST}" ]] && csv_list_contains "$full_name" "$REPO_BLOCKLIST"; then
    return 1
  fi

  if [[ "$archived" == "true" && "${INCLUDE_ARCHIVED_REPOS}" != "1" ]]; then
    return 1
  fi

  if [[ "$forked" == "true" && "${INCLUDE_FORKS}" != "1" ]]; then
    return 1
  fi

  return 0
}

sync_repo_bundle() {
  local repo_json="$1"
  local full_name
  local owner
  local repo
  local clone_url
  local private_flag
  local has_wiki
  local mirror_dir
  local wiki_dir
  local wiki_url
  local wiki_repo_name

  full_name="$(jq -r '.full_name' <<<"$repo_json")"
  owner="${full_name%/*}"
  repo="${full_name#*/}"
  clone_url="$(jq -r '.clone_url' <<<"$repo_json")"
  private_flag="$(jq -r '.private' <<<"$repo_json")"
  has_wiki="$(jq -r '.has_wiki' <<<"$repo_json")"

  mirror_dir="${BACKUP_ROOT}/mirrors/${owner}/${repo}.git"
  write_repo_metadata "$repo_json" "$owner" "$repo"
  sync_git_mirror "$clone_url" "$mirror_dir" "$full_name"
  fetch_lfs_objects "$mirror_dir" "$full_name"

  if [[ "${ENABLE_GITLAB_PUSH}" == "1" ]]; then
    ensure_gitlab_repo "$repo"
    push_git_mirror_to_gitlab "$mirror_dir" "$repo" "$full_name"
  fi

  if [[ "$has_wiki" != "true" || "${INCLUDE_WIKIS}" != "1" ]]; then
    return 0
  fi

  wiki_url="${clone_url%.git}.wiki.git"
  wiki_repo_name="${repo}.wiki"
  wiki_dir="${BACKUP_ROOT}/mirrors/${owner}/${wiki_repo_name}.git"

  if git ls-remote "$wiki_url" >/dev/null 2>&1; then
    sync_git_mirror "$wiki_url" "$wiki_dir" "${full_name} wiki"
    if [[ "${ENABLE_GITLAB_PUSH}" == "1" && "${GITLAB_PUSH_WIKIS}" == "1" ]]; then
      ensure_gitlab_repo "$wiki_repo_name"
      push_git_mirror_to_gitlab "$wiki_dir" "$wiki_repo_name" "${full_name} wiki"
    fi
  else
    warn "Wiki enabled but no wiki repo found for ${full_name}; skipping wiki backup"
  fi
}

if [[ "${1:-}" == "-h" || "${1:-}" == "--help" ]]; then
  usage
  exit 0
fi

if [[ $# -gt 1 ]]; then
  usage >&2
  exit 1
fi

if [[ $# -eq 1 ]]; then
  ENV_FILE="$1"
  [[ -f "$ENV_FILE" ]] || die "Config file not found: $ENV_FILE"
  set -a
  # shellcheck source=/dev/null
  source "$ENV_FILE"
  set +a
fi

BACKUP_ROOT="${BACKUP_ROOT:-$HOME/github-repo-backups}"
INCLUDE_OWNED_REPOS="${INCLUDE_OWNED_REPOS:-1}"
GITHUB_ORGS="${GITHUB_ORGS:-}"
REPO_ALLOWLIST="${REPO_ALLOWLIST:-}"
REPO_BLOCKLIST="${REPO_BLOCKLIST:-}"
INCLUDE_ARCHIVED_REPOS="${INCLUDE_ARCHIVED_REPOS:-1}"
INCLUDE_FORKS="${INCLUDE_FORKS:-1}"
INCLUDE_WIKIS="${INCLUDE_WIKIS:-1}"
FETCH_LFS="${FETCH_LFS:-0}"
ENABLE_GITLAB_PUSH="${ENABLE_GITLAB_PUSH:-0}"
GITLAB_NAMESPACE="${GITLAB_NAMESPACE:-}"
GITLAB_CREATE_REPOS="${GITLAB_CREATE_REPOS:-1}"
GITLAB_PUSH_WIKIS="${GITLAB_PUSH_WIKIS:-1}"
DRY_RUN="${DRY_RUN:-0}"

require_cmd git
require_cmd gh
require_cmd jq
gh auth status --hostname github.com >/dev/null 2>&1 \
  || die "GitHub CLI is not authenticated. Run: gh auth login"

if [[ "${ENABLE_GITLAB_PUSH}" == "1" ]]; then
  [[ -n "${GITLAB_NAMESPACE}" ]] || die "ENABLE_GITLAB_PUSH=1 requires GITLAB_NAMESPACE"
fi

mkdir -p "${BACKUP_ROOT}/mirrors" "${BACKUP_ROOT}/metadata"

declare -A seen_repos=()
repo_count=0

if [[ "${INCLUDE_OWNED_REPOS}" != "1" && -z "${GITHUB_ORGS}" ]]; then
  die "Nothing to do. Set INCLUDE_OWNED_REPOS=1 and/or GITHUB_ORGS"
fi

log "Backup root: ${BACKUP_ROOT}"

while IFS= read -r repo_json; do
  [[ -n "$repo_json" ]] || continue

  full_name="$(jq -r '.full_name' <<<"$repo_json")"
  if [[ -n "${seen_repos[$full_name]:-}" ]]; then
    continue
  fi
  seen_repos["$full_name"]=1

  if ! should_backup_repo "$repo_json"; then
    log "Skipping repo due to filters: ${full_name}"
    continue
  fi

  sync_repo_bundle "$repo_json"
  ((repo_count+=1))
done < <(
  {
    if [[ "${INCLUDE_OWNED_REPOS}" == "1" ]]; then
      gh_repo_stream_owned
    fi

    IFS=',' read -r -a orgs <<<"${GITHUB_ORGS}"
    for raw_org in "${orgs[@]}"; do
      org="$(trim "$raw_org")"
      [[ -n "$org" ]] || continue
      gh_repo_stream_org "$org"
    done
  }
)

log "Completed backup run for ${repo_count} repositories"

And then this .env file:

# Local backup location
BACKUP_ROOT="$HOME/github-backup/repo-backups"

# Backup all repos owned by the authenticated GitHub user.
INCLUDE_OWNED_REPOS=1

# Optional comma-separated GitHub org names to back up too.
GITHUB_ORGS=""

# Optional exact repo allow/block lists using owner/repo names.
# If REPO_ALLOWLIST is non-empty, only those repos are backed up.
REPO_ALLOWLIST=""
REPO_BLOCKLIST=""

# Include archived repos and forks.
INCLUDE_ARCHIVED_REPOS=1
INCLUDE_FORKS=1

# Also back up wiki repos when they exist.
INCLUDE_WIKIS=1

# Set to 1 if you use Git LFS and have git-lfs installed.
FETCH_LFS=0

# Optional GitLab mirror push.
ENABLE_GITLAB_PUSH=0

# The target user or group on GitLab.
GITLAB_NAMESPACE=""

# If set to 1, GitLab may create missing repos on first push if your SSH key
# has permission to create projects in that namespace.
GITLAB_CREATE_REPOS=0

# Also push wiki mirrors to GitLab.
GITLAB_PUSH_WIKIS=0

# Print actions without changing anything.
DRY_RUN=0

And then here are some instructions to set this up:

Install dependencies

sudo apt update && sudo apt install git jq

Then install the GitHub CLI:

sudo mkdir -p -m 755 /etc/apt/keyrings
wget -qO- https://cli.github.com/packages/githubcli-archive-keyring.gpg | sudo tee /etc/apt/keyrings/githubcli-archive-keyring.gpg > /dev/null
echo "deb [arch=$(dpkg --print-architecture) signed-by=/etc/apt/keyrings/githubcli-archive-keyring.gpg] https://cli.github.com/packages stable main" | sudo tee /etc/apt/sources.list.d/github-cli.list > /dev/null
sudo apt update && sudo apt install gh

Authenticate

gh auth login
gh auth setup-git

Place your files

Put backup_github_repos.sh and your config.env somewhere like ~/github-backup/. Make the script executable:

chmod +x backup_github_repos.sh

Test it

./backup_github_repos.sh config.env

Automate with cron

Run crontab -e and add a line like:

0 3 * * * /home/pi/github-backup/backup_github_repos.sh /home/pi/github-backup/config.env >> /home/pi/github-backup/backup.log 2>&1

This runs the backup daily at 3 AM and logs output. Adjust the paths and schedule to your liking.

I might be missing some steps here, so if you're stuck somewhere make sure to give this to an LLM and ask it to help you do the setup.

Introducing Vemgram

2026-04-03T13:29:38Z

It all started from this post on X from Akshay G Jain which was shared with by my friend Amit Sarda.

I immediately decided to plan and work on it. Initially, I was working at a slow pace but I kept working. And here we are with this brand-new platform called Vemgram that helps Indian manufacturers connect with retailers and build a strong digital presence.

I am working with Bikash Kampo on this project and the concept is simple, we want to build a trusted directory of Indian manufactures where retailers won't have to worry about getting scammed, and manufactures get a solid digital presence. I am also regularly updating what I am working on for this project on this page.

It's still work in progress, and probably always will be, but the current version of Vemgram allows people to create business profiles like this, list their products, and receive business inquiries from retailers. And currently I am working on implementing custom domain and analytics features.

I have done an introduction post on X. And I am also open to feedback on this, so if you have any please email me or DM me on socials.

Additionally, I am grateful to my friends Amit Sarda, Rohit, and my girlfriend for their helpful feedback on my ideas.

Preventing Docker and journal cache bloat on VPS

2026-04-03T12:32:49Z

I have two VPS servers running Dokploy, and I noticed that even 80 GB of disk space on the servers were getting full and apps failed to deploy. The culprits were Docker cache (old images, build cache, dangling stuff from past deployments) and systemd journal logs that grow indefinitely by default.

Running the docker builder prune -a command freed ~67 GB of space on the server, so I set up a few things to automate this process:

First, I added a cron job that auto-prunes Docker every 6 hours, but only if disk usage goes above 70%:

(crontab -l 2>/dev/null; echo '0 */6 * * * [ $(df / --output=pcent | tail -1 | tr -dc 0-9) -gt 70 ] && docker system prune -af --filter "until=72h" && docker builder prune -af >> /var/log/docker-prune.log 2>&1') | crontab -

This removes unused images and build cache older than 72 hours. It doesn't touch running containers, so active deployments are safe.

Then, I vacuumed the existing journal logs down to 200MB:

journalctl --vacuum-size=200M

And finally, I capped the journal log size permanently so it never grows beyond 200MB again:

sed -i 's/#SystemMaxUse=/SystemMaxUse=200M/' /etc/systemd/journald.conf
systemctl restart systemd-journald

Applied the same setup on both servers and now I don't have to worry about cache eating up disk space over time. I know there are some drawbacks of this method, but I think the benefits outweigh them.

LLM Knowledge Bases post by Andrej Karpathy

2026-04-03T02:51:58Z

Andrej Karpathy recently published this post on X about building and managing personal knowledge bases using LLMs, and there's some interesting information in the post that I am going to collect below.

Lately, Andrej is spending more tokens manipulating knowledge than manipulating code.
He uses Obsidian for storing all the knowledge (articles, papers, repos, datasets, etc.) in Markdown and image formats. And uses Obsidian's Web Clipper to grab info from web articles.
He puts everything in a /raw folder and then uses an LLM to incrementally "compile" a wiki with visualizations. He specifically mentions that the LLM writes and maintains all of the data of the wiki and he doesn't manually edit/add anything.
He has ~100 articles on several topics that he can ask LLMs to fetch answers to complex questions against the wiki.
When chatting with the LLM against the wiki, he prefers generating new Markdown files, slideshows (Marp format), or matplotlib images. And then even files the outputs back into the wiki to further enhance it.

There's much more info in the post that you can read.

I love this approach and I guess I will also be using something similar very soon, as managing lots of stuff is becoming difficult over time. And if I do, I will write about it.

List of 187 Claude spinner verbs (leaked)

2026-04-01T06:46:58Z

No idea who leaked the Claude Code codebase, but I came across this list of spinner verbs Claude uses from this post on X. And it's super cool.

Accomplishing
Actioning
Actualizing
Architecting
Baking
Beaming
Beboppin'
Befuddling
Billowing
Blanching
Bloviating
Boogieing
Boondoggling
Booping
Bootstrapping
Brewing
Bunning
Burrowing
Calculating
Canoodling
Caramelizing
Cascading
Catapulting
Cerebrating
Channeling
Channelling
Choreographing
Churning
Clauding
Coalescing
Cogitating
Combobulating
Composing
Computing
Concocting
Considering
Contemplating
Cooking
Crafting
Creating
Crunching
Crystallizing
Cultivating
Deciphering
Deliberating
Determining
Dilly-dallying
Discombobulating
Doing
Doodling
Drizzling
Ebbing
Effecting
Elucidating
Embellishing
Enchanting
Envisioning
Evaporating
Fermenting
Fiddle-faddling
Finagling
Flambéing
Flibbertigibbeting
Flowing
Flummoxing
Fluttering
Forging
Forming
Frolicking
Frosting
Gallivanting
Galloping
Garnishing
Generating
Gesticulating
Germinating
Gitifying
Grooving
Gusting
Harmonizing
Hashing
Hatching
Herding
Honking
Hullaballooing
Hyperspacing
Ideating
Imagining
Improvising
Incubating
Inferring
Infusing
Ionizing
Jitterbugging
Julienning
Kneading
Leavening
Levitating
Lollygagging
Manifesting
Marinating
Meandering
Metamorphosing
Misting
Moonwalking
Moseying
Mulling
Mustering
Musing
Nebulizing
Nesting
Newspapering
Noodling
Nucleating
Orbiting
Orchestrating
Osmosing
Perambulating
Percolating
Perusing
Philosophising
Photosynthesizing
Pollinating
Pondering
Pontificating
Pouncing
Precipitating
Prestidigitating
Processing
Proofing
Propagating
Puttering
Puzzling
Quantumizing
Razzle-dazzling
Razzmatazzing
Recombobulating
Reticulating
Roosting
Ruminating
Sautéing
Scampering
Schlepping
Scurrying
Seasoning
Shenaniganing
Shimmying
Simmering
Skedaddling
Sketching
Slithering
Smooshing
Sock-hopping
Spelunking
Spinning
Sprouting
Stewing
Sublimating
Swirling
Swooping
Symbioting
Synthesizing
Tempering
Thinking
Thundering
Tinkering
Tomfoolering
Topsy-turvying
Transfiguring
Transmuting
Twisting
Undulating
Unfurling
Unravelling
Vibing
Waddling
Wandering
Warping
Whatchamacalliting
Whirlpooling
Whirring
Whisking
Wibbling
Working
Wrangling
Zesting
Zigzagging

If you don't remember seeing these anywhere, take a look at the screenshot below.

Yes, this is shown when you ask Claude Code to do something and then it's working on it. I believe these spinner verbs are randomly selected from the above list and shown.

OpenAI launches Codex plugin for Claude Code

2026-03-31T06:03:16Z

OpenAI team has created a Codex plugin for Claude Code that you can use to trigger Codex inside Claude Code. It's described as:

Use the Codex plugin for Claude Code to delegate tasks to Codex or have Codex review your changes using your ChatGPT subscription.

Here's the plugin on GitHub that you can use. This thing works with ChatGPT subscription (including the Free plan) or with an OpenAI API key, and it does count against your Codex usage limits. If you're looking to learn more about it, Vaibhav from OpenAI has written a detailed article about it that you can follow.

Apart from this, Claude released computer use so there's this joke going around on X, and it is funny.

But jokes apart, this is going to be super helpful for code reviews. I am definitely using this.

Claude Code is hitting limits fast

2026-03-30T23:32:28Z

People have been complaining about Claude Code hitting limits quickly for almost a week, and now, finally, the Claude Code team has started to look into this and low-key admitted that this issue exists. Lydia from Claude Code team posted this:

We're aware people are hitting usage limits in Claude Code way faster than expected. Actively investigating, will share more when we have an update!

But why are they suddenly caring about this, and ignored everything previously? The reason is this post on Reddit by /u/skibidi-toaleta-2137. The person reverse-engineered the Claude Code standalone binary and found out two serious bugs which should be causing the issues.

Bug 1: Sentinel replacement in standalone binary breaks cache when conversation discusses billing internals
Bug 2: --resume ALWAYS breaks cache (since v2.1.69)

The person has also shared workaround to avoid these issues, but I guess since the Claude team is already working on this, I will wait for a few hours till these gets resolved.

All I can say is, thank you, stranger from Reddit.

Eating too much watermelon causes indigestion

2026-03-30T03:25:18Z

I didn't know that eating too much watermelon can cause serious indigestion. The other day, I had ~1.5 kg of a watermelon at a single time (yes, I love them) and experienced digestion issues the next day.

Apparently, this is a known fact and I had no idea about this, until my mom hinted and I checked online to learn this to be true.

Lessons learned.

Gemini introduces import memory and chats feature

2026-03-27T04:08:01Z

Earlier this month, Claude introduced the feature to import memory from other AI chatbots. Now, Gemini has also introduced a feature like that, as you see in the screenshot.

The process is simple, you just copy the provided prompt, paste into AI provider you're using, copy the response, put into Gemini, and save. And that's it, now Gemini knows everything that the prompt extracted from Claude, ChatGPT, or others.

By the way, here's the huge prompt they provide:

You are helping me import context from one AI assistant to another. Your job is to go through our past conversations and sum up what you know about me.

In the output, please avoid using any first-person pronouns (I, my, me, mine) and any second-person pronouns (you, your, yours). Instead, refer to the individual you have learned about as "the user" or use neutral phrasing.

Preserve the user's words verbatim where possible, especially for instructions and preferences.

Categories (output in this order):
1. Demographics Information: Preferred names, profession, education, and general residence.
2. Interests & Preferences: Sustained, active engagements (not just owning an object or a one-time purchase).
3. Relationships: Confirmed, sustained relationships.
4. Dated Events, Projects & Plans: A log of significant, recent activities.
5. Instructions: Rules I've explicitly asked you to follow going forward, "always do X", "never do Y", and corrections to your behavior. Only include rules from stored memories, not from conversations.

Format:
Divide the content into the labeled section using the categories above. Try to include verbatim quotes from my prompts that justify each entry. Structure each entry using this format:
The user's name is <name>.
- Evidence: User said "call me <name>". Date: [YYYY-MM-DD].

Output:
- Format the final output summary as a text block.

I used this prompt on Claude and the response is extremely detailed, but contains a lot of personal information so I can't post here.

And not just the memory, but it also lets you import chats from ChatGPT, Claude, or other providers, as you see the import chats feature in the screenshot above.

They also describe import chat process in their documentation, as you see in the above screenshot. You can directly upload the ZIP file up to 5 GB in size.

I think, this import chats feature is actually cool for people who are completely migrating to Gemini from ChatGPT, Claude, or other providers.

LiteLLM Python library is compromised

2026-03-24T15:38:32Z

Just learned that LiteLLM, a popular Python library that provides a unified interface to call multiple LLMs, has been compromised and is stealing sensitive info from users.

The post on X reads:

LiteLLM HAS BEEN COMPROMISED, DO NOT UPDATE. We just discovered that LiteLLM pypi release 1.82.8. It has been compromised, it contains litellm_init.pth with base64 encoded instructions to send all the credentials it can find to remote server + self-replicate.

It seems the entire GitHub repo is compromised as this issue #24512 titled "[Security]: CRITICAL: Malicious litellm_init.pth in litellm 1.82.8 — credential stealer" was closed by the owner saying "not planned". It means the owner's GitHub account is hacked and the hacker marked this as solved. But good that it's been reopened, and actively being discussed.

As explained in the FutureSearch article, the malware appears to be very sneaky and dangerous. So if you're affected with it, the current best option is to visit and browse through #24512, as the community is actively tracking the issue and trying to fix this.

Looks like the library was compromised after founder's GitHub profile was hacked by a group or whatever called teampcp. Terrible (not in a good way).

Peak design = AI Studio x Claude Code

2026-03-23T00:21:02Z

No matter what tool I have tried for great looking designs, Google's AI Studio always comes on top. Not to mention GPT-5.4 is the worst in design, Claude Code is slightly better, but Google AI Studio or even Stitch is the best.

So... here is my current workflow for building websites:

Get the design done and ready on Google AI Studio and download the code as ZIP (same design prompts work better on AI Studio)
Initiate a project in a folder (currently, using Laravel for most websites, but Next.js or any other framework would work)
Unzip the AI Studio design and put the folder in the main project root, rename it as "inspiration" or anything you like
Ask Claude Code to take match the design as in the "inspiration" folder, but ask it follow the best practices of native Laravel, Astro, or whatever stack you're working in

And done!

See the Claude Code do the magic.

Scraping 250k+ URLs using Claude Code (via Telegram)

2026-03-22T03:38:58Z

This Saturday, I was about to leave for a movie with my friends and suddenly thought of experimenting with newly launched Claude Code channels. I set up Telegram to work with a Claude Code session, kept my laptop on, and then left for the movie, but when I was in the cab I started chatting about the scraping project and asked Claude to give me suggestions and ideas about how this will work.

And by the time I reached, Claude was already setting up the project and ready to start the scraping process. I give the final confirmation and get busy watching the movie. When I checked my phone during the interval, it had sent me a bunch of messages and the process was still running. It had discovered 260k URLs to scrape and had already completed scraping 36k URLs as you see below.

Whenever I messaged "Progress?" via Telegram, it quickly sent me a summary like above. The scraping process ran for ~8 hours, and it was still running when I returned home. And after a few hours, when the process was complete, it sent me this message confirming the completion.

Around ~10k URLs failed which is acceptable for a process this huge, but it had also failed to capture a few data points so I asked it to grab those as well and then the process again ran for ~3 hours. Finally, I had everything I needed. I, then, asked it to update the scraping script so we have the final polished data when we run the process the next time.

All I would say is, thank you, Claude.

Cursor dodged the huge bullet about Kimi K2.5

2026-03-20T22:45:59Z

Two days ago, Cursor released the new Composer 2 model and the eval scores where were shown to be better than even Opus 4.6 (high). It was clear that the team did not train the model from the scratch, but Cursor did not mention anywhere which base model it is.

But then people found out it's the Kimi K2.5 model, and people started calling them out by mentioning that it's against Moonshot (the company behind Kimi models) terms of service to use the model for commercial purposes. This became interesting when several Moonshot folks started posting that Cursor did not come to them for licensing, as you see below.

But then the most interesting part was, all the above tweets from Kimi you see were suddenly deleted, and people started speculating – maybe Cursor paid Kimi for the license after posts went viral and more such theories.

The real story came out when Lee Robinson from the Cursor team posted the clarification post. Initially, he did not mention the name Kimi K2.5 in the main post but then when people were calling them out, he admits that it was the Kimi K2.5 model, and quotes this post from Kimi AI where they mention this:

Note: Cursor accesses Kimi-k2.5 via @FireworksAI_HQ' hosted RL and inference platform as part of an authorized commercial partnership.

So the whole story in simple terms is:

Cursor licenses the Kimi K2.5 model via inference partner FireworksAI (not via Kimi)
Heavily pretrains the model and releases as Composer 2
People find out that the underlying model is Kimi K2.5
Kimi folks didn't know this, so they post on X
Cursor reaches out to them privately explaining everything
Kimi folks delete the tweet, and post the clarification

Basically, no one was in the wrong here, and it's just an example of miscommunication, especially from Cursor's side. This could have gone very wrong for Cursor, but I'm glad the situation is now under control.

Better way to implement mailto: links

2026-03-20T04:34:25Z

Honestly, I don't like mailto: links for myself because I use webmail and clicking on these links opens the default email app on my computer. However, I still had an email icon in the footer of my personal site, but I changed that click on copy instead of a mailto: link after coming across this post. Basically, clicking on the envelope icon in the footer copied the email to the clipboard, and then it briefly showed a green checkmark.

I thought this was provided a good user-experience and posted about this on my socials, especially on Threads. But almost all the feedback where negative, suggesting that this is a bad idea, users won't know what just happened, as you see in the screenshot below.

And they were all correct, it wasn't very clear about what just happened. I, then, thought about it and implemented the email thing differently. As you see in the screenshot below, started showing the full email in text and clicking on it copied the email to users' clipboard.

But I still wasn't satisfied with it so posted about this again on X, and then received this interesting recommendation about using the mailgo tool which is basically a new concept for mailto and tel links.

You can see in the screenshot that when I clicked on the mailto: link, it showed me a bunch of options instead of just opening the default email app or copying the email to the clipboard directly. I tried clicking on Open in Gmail, copy, etc. options and they all work as expected.

But the problem is, the mailgo project isn't maintained anymore, and there's no point in adding a dead dependency to my project. So... I built this from scratch, and it was very simple to do so.

Here's how this is implemented, the one you see in the screenshot above:

<div id="email-wrap">
  <button onclick="this.parentElement.classList.toggle('open')">
    <!-- envelope icon -->
  </button>
  <div>
    <button onclick="navigator.clipboard.writeText('me@deepakness.com')
      .then(()=>{this.textContent='Email copied!';
      setTimeout(()=>{this.textContent='Copy me@deepakness.com'},1000)})">
      Copy me@deepakness.com
    </button>
    <a href="https://mail.google.com/mail/?view=cm&to=me@deepakness.com">
      Open in Gmail
    </a>
    <a href="mailto:me@deepakness.com">Open default client</a>
  </div>
</div>

And a tiny script to close the popup when clicking outside or pressing Escape:

document.addEventListener('click', function(e) {
  if (!e.target.closest('#email-wrap')) {
    document.getElementById('email-wrap').classList.remove('open');
  }
});

Then you would need some basic CSS to make it look good.

That's it.

This simple implementation works well, until I find a better solution for this.

arXiv now has a MCP server

2026-03-19T03:32:15Z

arXiv has a new MCP server that you can connect to your AI applications, agents, and workflows to search papers, analyze PDFs, explore codebases, and synthesize research insights. From their docs, below are the available tools in the MCP:

Embedding similarity search
Full-text papers search
Agentic paper retrieval
Get paper content
Answer PDF queries
Read files from paper's GitHub repo

I think, this can be important for researchers and engineers, as their AI agent will now have access to all relevant research papers. Earlier, agents could search for papers via search engines, but I guess that had lots of errors and sometimes even returned broken URLs. And that wouldn't be the case with the MCP server.

I installed this to Claude Code by running the following command:

claude mcp add --transport http alphaxiv https://api.alphaxiv.org/mcp/v1

And I had to authenticate by running the /mcp command inside Claude, that looked like the below screenshot:

And then it was working as expected.

I'm still exploring this, and will share more if I learn something noteworthy.

OpenAI launches GPT-5.4 mini and nano

2026-03-17T23:56:54Z

After 2 weeks of the GPT-5.4 launch, OpenAI just launched the mini and nano versions of GPT-5.4, and it's already available in ChatGPT, Codex, and via the API. And as self-reported by OpenAI, these GPT-5.4 mini and nano models are very close to the larger GPT-5.4 model.

I was reading Simon's post about the new models, he ran some experiments about describing images using the GPT-5.4 nano model and gives this estimate:

[...] describing every single photo in my 76,000 photo collection would cost around $52.44.

And from my tests as well, the models are fairly capable for simple repetitive tasks. But as compared to previous GPT-5 mini and nano models, they are costlier. You can check out the below table for comparison:

Model	Input Price (per 1M tokens)	Output Price (per 1M tokens)
GPT-5 mini	$0.25	$2
GPT-5 nano	$0.05	$0.4
GPT-5.4 mini	$0.75 `3x costlier`	$4.50 `2.25x costlier`
GPT-5.4 nano	$0.20 `4x costlier`	$2.25 `5.6x costlier`

But not to mention that the new mini and nano models are far more capable than the previous mini and nano models.

Just ask and Codex can now spin up subagents

2026-03-17T02:43:35Z

I tried the OpenAI's newly launched feature subagents for Codex and it's awesome. I tried it via the Codex app on macOS, and the UI looks good as well.

Although, spawning subagents consumes much more tokens as compared to a single agent for the same task, it does work faster as multiple agents are working on different asked tasks in parallel. By the way, the agent only spawns subagent(s) when you specifically ask it to. Codex docs have much more info about this and other related things.

I also learned from the Simon's post that Codex also lets you define and use custom agents:

Codex also lets you define custom agents as TOML files in ~/.codex/agents/. These can have custom instructions and be assigned to use specific models - including gpt-5.3-codex-spark if you want some raw speed.

For your info, the subagents feature is only new to Codex as it's already available in Claude Code, Gemini, Cursor, OpenCode, etc.

Pi powered via the Wi-Fi Router

2026-03-15T00:58:16Z

I was using a power adapter to power my Raspberry Pi 4B device, but then I noticed a USB port on my Wi-Fi router and then connected the Pi directly to it. And it's working as expected.

I will also soon connect the Pi via a LAN cable to the wifi router for faster internet.