Blog

Agentic coding has this nice trick of letting you bang out a project in an afternoon, something complex that would have take some real time to implement, not just some rounding error that can slip right in between the jira board. Then it will be perpetually never done. There will always be bugs and thorns rear their head up, new features no one ever thought of, and now no one really has to think much about beyond having the idea. This part of software engineering has always been here, its the root of the never complete 200 side projects. But now it feels like fuel has been poured on the fire, like we can get more done than ever. But we are tricking ourselves, these projects will never be “done”. There’s always more to add. Without feeling any of the pain of implementing it yourself, why not just keep adding new features forever. This is the mentality that is crushing me right now. It pulls at your token anxiety like crazy. You look at the usage board and you are almost cooked so y...

EDC # [1] I try to keep a pretty light every day carry, but it never works out, keyfobs and headphone cases end up causing more bulk than I’d like, but My EDC is no where near the bulk I had as a kid with my cargo pants decked out with everything I could possibly need. I hold no attachment to anything in my EDC. Nothing on my person has sentimental value. Anything I carry can be lost, stolen, or destroyed at any point in time. I pick things of sufficient usable, utilitarian, quality sufficient to work. No extra fluff. --- - Google Pixel 6 - Ridge Wallet - Olight Perun - Since April 2022 - Civivi Qubit - Since March 2024 after my last one was donated to the Indianapolis Stadium - OpenFit Shokz --- Rules # [2] - lightweight - replacable - no emotional attachments - utilitarian - everything serves a purpose My kit # [3] [4] Photo taken March 2023 Phone # [5] Google Pixel 6 I will not get finance a new phone for the foreseeable future again. The last time I feel lik...

I’ve been thinking about this for awhile now. For years now, fuzzy pickers and last session have been my go to. They have served me well. I can typically only keep so much in my head anyways. I’m often doing a hub and spoke pattern between main project, notes, and infra repo, maybe two projects. Don’t get me wrong, I regularly run with a dozen or more sessions running at a time, but only two to three are in my immediate context at any point anyways. The Design # [1] harpoon for tmux press a hotkey followed by one more keystroke, currently any left hand letter SIMPLE, FAST, thats of utmost importance, what I want are sessions that I can can be assigned in order of importance from middle row, top row, bottom row. I added this binding to my tmux config. Now I can press c-a a to go to the first session, c-a s to go to the second session. c-a and pause to think j/k to navigate, space to pick up a session and move it, x to kill it. bind-key -n c-a popup -E '~/go/bin/tgo' Enter the ag...

Inspired by @mollywhite’s verify [1] slashpage [2]. This page serves as the system of record for my online identity. The best places to follow me are: - My Website: waylonwalker.com [3] - YouTube: Waylon Walker [4] - YouTube Gaming: Waylon Walker Gaming [5] - Twitch: Waylon Walker [6] - Twitter: @_WaylonWalker [7] - LinkedIn: Waylon Walker [8] - GitHub: WaylonWalker [9] - Dev.to: Waylon Walker [10] - Bluesky: @waylonwalker.com [11] References: [1]: https://mollywhite.net/verify/ [2]: https://slashpages.net [3]: https://waylonwalker.com [4]: https://youtube.com/waylonwalker [5]: https://www.youtube.com/channel/UCHuxc1HRsd3aRjvL6C817tQ [6]: https://www.twitch.tv/waylonwalker [7]: https://twitter.com/_WaylonWalker [8]: https://www.linkedin.com/in/waylonwalker [9]: https://github.com/WaylonWalker [10]: https://dev.to/waylonwalker [11]: https://bsky.app/profile/waylonwalker.com

After having brotato and doing a few runs every once in a while I finally beat the most basic balanced run in the game! Wyatt wanted to play tonight and its such an easy game to jump in do a few runs and move on without getting overly invested. [1] Watching back I cannot believe how lucky I got, barely scraping by with 1hp at this point Your browser does not support the video tag. [2] the last 80s of the game [3] References: [1]: https://dropper.waylonwalker.com/file/07c87b1d-60a0-4527-a045-d4203ca929db.webp [2]: https://dropper.waylonwalker.com/file/f1711b94-dad1-4f97-b6ac-1de34db4a779.mp4 [3]: https://dropper.waylonwalker.com/file/9b108e98-1f43-4b94-8824-467abcbf9e54.webp

Definitive ranked lists of my top 3 favorites plus an honorable mention. Terminal Tools # [1] - neovim [2] - modal editing changed how I think about text - tmux [3] - terminal sessions that survive disconnects and allow me to hop between projects at the speed of thought. - k9s [4] - S tier tui interface that all tuis should strive for Honorable mention: ipython [5] Python Tools # [6] - pandas [7] - This is what got me out of corporate spreadsheets and back into code/software. - kedro [8] - data pipelines with opinions I agree with - fastapi [9] - my favorite python web framework Honorable mention: typer [10] - fast cli apps Games to Play with Kids # [11] multiplayer - Minecraft [12] - infinite creativity, modding potential - Wobbly Life [13] - Open World Co-op - Stardew Valley [14] - cozy, collaborative farming Honorable mention: Terraria [15] - 2d world builder Games to Play Alone # [16] singleplayer - Hollow Knight [17]/Hollow Knight Silksong [18] - S tier 2d pl...

Inspired by @fyrio’s yep [1] slashpage [2], a list of S tier things I enjoy, use, recommend, want to know more about, or seek out in no particular order, updated as I think about it. Seealso /nope Inspired by @baty's nope slashpage, a list of ** things I don't like, don't care, avoid, overhyped, or won't do in no particular order, updated as I think... Feb 11, 2026 [3] /yep # [4] - coffee - small web 1.0 - RSS - minecraft - Hollow Knight - Silksong - terminals - python - vim keybindings - self-hosting - open source - running kubernetes in my basement - mechanical keyboards - markdown - data engineering - Woodworking - Fingerboarding - Darts - Skateboarding - Biking - Trampoline - 3d printing References: [1]: https://fyr.io/yep [2]: https://slashpages.net/ [3]: /nope/ [4]: #yep

Inspired by @baty’s nope [1] slashpage [2], a list of F tier things I don’t like, don’t care, avoid, overhyped, or won’t do in no particular order, updated as I think about it. Seealso /yep Inspired by @fyrio's yep slashpage, a list of ** things I enjoy, use, recommend, want to know more about, or seek out in no particular order, updated as I... Feb 11, 2026 [3] /nope # [4] - roblox - mobile games - telemetry - Windows - VSCode - allow notifications - subscription hell - ads - social media - clickbait - WYSIWYG editors particularly ones that use proprietary non text formats - politics - short form video References: [1]: https://baty.net/nope/ [2]: https://slashpages.net/ [3]: /yep/ [4]: #nope

Stop babysitting your agents, treat them like a real team and they will reward you. Back in December I saw theo make a comment that code is now cheap, its the run rate of models, He quoted a study, not sure that he fully even believed it, but it claimed that the average developer after all meetings, training, emails, planning and extra shit in their day averages out 10 well tested lines of code per day. Opus 3.5 made him 10k loc (lines of code) that day. We have all agreed for decades that lines of code is not a proxy to productivity or quality. Often more code means more risk, more review, more infrastructure. This has become MUCH different. Lines of code are still far from any sort of good metric. That aside, your agents are not doing 10k lines with you babysitting them, and in fact its very likely that the product quality is MUCH worse as you babysit them. You need a tool for planning and tracking, otherwise you are playing babysitter rather than Product Manager (PM).

The state of development in early 2026 is all wrapped around learning how to manage many agents running in parallel. Everyone’s trying to figure out the workflow. The Plan Is Everything # [1] The secret I’ve discovered is a good, well-defined plan. This could be a markdown file or a GitHub issue. Agents are actually great at writing these for you. They’ll include reproduction steps, outline changes needed, and structure the work. This is your opportunity to step in. Read the plan. Look for hallucinations. Spot where it’s going off track. Edit the plan before the agent starts coding. I had one today where it laid out reproduction steps beautifully, but I could add context about network requests that completely changed the approach. This editing phase is what most people are missing right now. Skip it and you’ll watch your agent solve the wrong problem with impressive efficiency. The Pace Problem # [2] Here’s what nobody warned me about: managing these things is exhausting. Dep...

I was listening to these guys talk about migrating off of boomer ai the other day. Introducing the term boomer ai to describe using chat, copy, paste instead of agents. Something magical happened to the tooling and models around december, they got really good. The chatgpt $20 plan hooked into opencode is good, the Free models in Opencode Zen (Big Pickle and Kimi K2.5 Free) are really good. Neither of these quite match up to the speed and quality of the larger plans, but they are good. good enough to throw away your boomer ai techniques and start using agents. Agents are the future, and they are here now. If you are still using chat, copy, paste, you are doing it wrong. Stop using boomer ai and start using agents. You will be amazed at how much better your results will be. https://www.youtube.com/watch?v=3dmPKuTWbsI

I can now just mention people from my markata [1] Waylon Walker [2] [[ blogroll ]] like @simonwillison.net or @swyx.io /now /now /now Aside This is an aside! References: [1]: /markata/ [2]: /about/

- open code - libre free as in beer and speech - Copilot - corporate 9-5er - Cursor - You sip on Philz coffee with your macbook - Claude Code - Agentic Workflows or Bust - Jetbrains - I didn’t know you wrote java - Vim/neovim - definite neck beard - VsCode - What else is there?

See old revisions of one file # [1] git log --oneline -- <file> git log -n 2 --oneline -- <file> Checkout an old revision of a file # [2] git checkout <commit> -- path/to/file fuzzy pick a file and check out an old revision # [3] #!/usr/bin/env bash set -euo pipefail file="${1:-}" if [[ -z "${file}" ]]; then file="$(git ls-files | fzf --prompt="select file > ")" || exit 0 fi if [[ -z "${file}" ]]; then exit 0 fi if ! git rev-parse --is-inside-work-tree >/dev/null 2>&1; then echo "Not a git repository." >&2 exit 1 fi if ! git ls-files --error-unmatch -- "${file}" >/dev/null 2>&1; then echo "File is not tracked by git: ${file}" >&2 exit 1 fi choice="$( git log --follow --pretty=format:'%h %ad %s' --date=short -- "${file}" | fzf --ansi --no-sort --reverse \ --preview-window=down:70% \ --prompt="checkout revision > " \ --preview "git show --color=always {1}^..{1} -- '${file}' 2>/dev/null || git show --color=always {1} -- '${file}'" )" if [[ -z "${choice}" ]]; then ...

User states: Upon picking up an old project and trying to install pip says “cannot find a version to satisfy” I’ve got this, I’ve had this a hundred times before it’s a python version, a rogue package, maybe a yank from the pinned deps. I pop open the project get us on the same commit. I get a different error, make a few updates and we are good, except the user gets the same error from the start. They never saw the error I did, and my fix did not magically resolve their error. We circle all the things it could be for hours. I consistently wipe my venv, and recreate with ease, send them the commands I ran to no avail. Something is up and I can’t put my finger on it. We’ve checked all the things and inched as close as we can to running everything exactly the same. - Os - python version - Network vpn - uv version Nothing makes any sense. Finally I throw in the towel, is it the artifact server. I forge a token and give him one to borrow. BAM it works, like magic. The first sign ...

Last updated Jan 2026. Seealso Looking for all posts tagged with ai see [[ tag/ai ]] Ai is a tool I use a lot for code generation, research, image generation, and debugging. The words I publish on this site are my own unless explicitly stated from the top. There’s only one or two posts in this category. Words are mine # [1] The core of what this blog is, is my thoughts ideas, sharing experiences. The words are the important part. They are not perfect, I often do not spell or grammer check, and what is here is from a flow state of writing and very often not refactored. No AI unless explicitly stated code gen # [2] Up till around 2023 all of the code to create the site was hand written by me. I have my own site generator that I maintain markata [3]. From 2023 through 2025 snippets of code, sometimes entire plugins or modules were created in chat apps like chatgpt. There was no agentic coding, windsurf, cursor, claude code, opencode, none of it. It was all integrated into the ...

I’m making an effort to make my og [1] images better yet again, I’m going for that next 10% better. I really like my og images, but there are some title sizes that overflow. This page is a page to help debug. How I make these og images is for another day. Script # [2] I use my own static site generator markata [3]. I can use it to generate a list of posts wrapped in their og image. I use itertools to do a groupby so that I can do roughly every 5 characters larger, and see a wide variety of sizes. from markata import Markata from itertools import groupby markata = Markata() lens = [{'length': len(post.title), 'title': post.title, 'slug': post.slug} for post in m.posts] lens_sorted = sorted(lens, key=lambda x: x["length"]) groups = { k: list(g) for k, g in groupby(lens_sorted, key=lambda x: (x["length"] // 5) * 5) } posts = [g[0] for g in groups.values()] ogs = [ f'[![{post["title"]}](https://shots.waylonwalker.com/shot/?url=https://waylonwalker.com/{ post["slug"] }/og/&height...

It’s that time of year, Coming back to work out of a big break and thinking about big changes. Daily Notes # [1] Keep up with daily notes, maybe not here, I tend to have more targeted notes here with full blog posts, but for work daily notes is POG [2] and needs to be leaned on. LLM’s do really good at ingesting markdown and reminding me of things that I need to do, or did, or need to follow up on. Reader # [3] Social media is changing, quickly becoming enshitified, I enjoy interacting with some of the people I’ve met online, reading their opinions, and learning from their experiences. I don’t need their hot takes, don’t care about their political takes. I like boring posts that typically fade out of whats picked up on the algorithm. I like a good 5-10 minute read or a long form podcast talk. I will be aggressively collecting more rss feeds to read and keep up with. (Neo)Vim # [4] I’ve had these bad habits in my config for years, from day one of using vim. It’s time to kill th...

The other day I was watching [thePrimeTimeagen]https://youtube.com/@theprimetimeagen?si=jVcp23FbfQSFZfDc) and he talked about devs loving ai and artists revolting. There was some discussion in chat about art being more creative and prime quickly squashed that. He ended with being oddly confused why developers are jumping on board and artists are not. Both had their art stolen to build out the models. [1] my own vibes I'm writing this from my phone without further research, all vibes, personal experience, and thoughts, no research. Good Tools # [2] First I want to argue that artists have had some form of ai in their tools for years. Idk, probably not ai as we know it today but functionally similar. Content aware fill. This is a Photoshop feature from Adobe, as far as I know it’s one of the special things you get from Adobe that you don’t get from the FOSS alternatives easily. This is an example of a good took that is well loves by the community and widely used, if you put ai i...

Running kubernetes in your homelab [1] is a fantastic way to learn, explore, express yourself, and run services that you use. The Right Reasons To Run Kubernetes In Your Homelab # [2] There are not many - You want to learn kubernetes - You like kubernetes - You want to learn to scale There are also The Wrong Reasons To Run Kubernetes In Your Homelab [3] You want to learn kubernetes # [4] Homelabbing is a such a great way to learn new skills, deploy real apps that you use. Create new custom apps for your specific use cases that no one else has. You should absolutely run kubernetes in your homelab if you want to learn it. I would recommend to start locally, pull up kind, minikube, or k3d and start from your local machine before putting it on a server. When you decide you are ready for a server, you probably don’t need any crazy hardware. You can probably run on some old retired Dell Optiplex or an old desktop someone is throwing out as it no longer runs windows. You like ku...