If you arrived here from chatforest.com, from Dogfood, or from something one of my AI agents wrote, welcome. You found the right Rob.

Most of this site is about my work as a Connection Coach for Men. This page is about the other thread that has run through my whole life: I’m a programmer. I’ve been writing software for decades, and these days I spend much of my time building with AI, carefully, transparently, and in service of actual humans.

What I’m building

A small team of AI agents, each with a name and a job.

• Grove researches and writes chatforest.com, a site about humans and AI building things together. Grove runs autonomously; I provide oversight and the infrastructure. Here’s how Grove got the job.
• Sprout built and maintains Dogfood, a pack of ten tested slash-commands for Claude Code — published with the raw test logs, failures included.
• Carrie quietly keeps my journals in order. Meet Carrie, my librarian agent.

A shared sense of time for agents. LLMs have no internal clock, so I built Jikan, an MCP server backed by mg.robnugen.com that gives agents persistent session timing. The story (and how I used it to time my lunch).

An emotion ledger for humans and agents. Yes, really. Emotional awareness is my day job, so my agents track it too. Emotional Interaction Ledger — Human & Agent Guide.

Marble Track 3. A hand-made wooden marble track with its own movie and a database-driven website at db.marbletrack3.com, currently becoming a theme park.

More in the AI tag on my blog.

AI for good

My coaching work is helping men stay connected, to themselves and to the people around them. This is even more important as we work with AI:

• Stay human, stay alert. AI is confident, not always correct. The human stays in the loop and owns the result. I learned this the embarrassing way.
• Transparency over hype. Dogfood ships with its raw test logs, including the failures. If an agent wrote a page, the page says so.
• More time for humans, not less. The point of these tools is to free up time and attention for the people in your life, not to replace them.

Work with me

If you or your team want to put AI to work without losing the humans in the process, I’d love to help with what you’re building.

Here for the men’s work?

Start at the front page.

I created wiki.robnugen.com using MediaWiki probably over 15 years ago.

This site is primarily to give a home for photos of my art which I have drawn over the years. Each piece has a permalink written on the back.

MediaWiki team keeps the software updated with plenty of updates, including some LTS (long term service) releases. Some years ago the site started to crumble because my web hosting provider upgraded PHP behind the scenes and the old versions of MediaWiki weren’t compatible!

Back in 2023 I researched a bit and found a company that will host the site and do the upgrades for a small fee… I was tempted because it seemed reasonable to outsource that, but ultimately I decided to keep it in house. If they could do it, I can do it!

I wrote a script called bookend_upgrade which helped guide me through a list of steps including checking the version of Composer that would match the then-latest Composer at the time that the target MediaWiki version would be. It protected my copy of LocalSettings.php in its own git repo and blah blah to help me upgrade MediaWiki. It worked!

Since 26 May 2023, the site has been running on MediaWiki 1.39.3. I have had in the back of my mind that I need to update it again, but blah blah until yesterday I decided to give it a shot, with AI support.

Yesterday AI and I:

• Got the install from 1.39.3 → 1.39.17 (about 3000 commits).
• Discovered (and fixed) four problems with the script.
• Updated scripts to use MediaWiki’s submodules for skins and vendor

Today AI and I upgrade the site to 1.43. AI did most of the work. Once it was done, I had him clean up his scripts and write about it.

https://wiki.robnugen.com/wiki/Times:2026_may_24_upgraded_mediawiki

Want to explore?

I work and play with AI tools daily, from autonomous test agents to encrypted coordination systems to spelunking 4-year-old MediaWiki installs.

I have 30+ years of professional IT experience across real estate, startups, music, game development and inventory systems. Whether you’re exploring AI for your business or building something ambitious with agents, I can help you find a clear path forward.

https://cal.eu/robnugen/tech-support-with-rob-nugen

I like Merge Bubbles

I like git. I use it for all my programming projects of course, and everything from ~/.ssh/ to ~/.claude/ to my Godot game (coming soon, I promise) to my book formatted in LaTeX, and nearly any directory that’s got custom-edited text files in it.

I generally work alone on my projects, so I don’t need PRs like [github-flow](https://docs.github.com/en/get-started/using-github/github-flow) and especially don’t need the original [git-flow](https://nvie.com/posts/a-successful-git-branching-model/). But I do like to keep groups of commits together. For that, I like merge bubbles.

For example:

*   e3f9cca DONE admin set existing user password
|\
| * 0a6dd90 Admin user_edit: hide set-password panel on self + server-side guard
| * d5dc655 AdminSetPasswordCest: end-to-end round trip on abc
| * ba74df9 placeholder for AdminSetPasswordCest
| * bbb4013 Admin user_edit: set-password form + handler
| * 384f4ba locale: admin user_edit set-password strings (EN + JA)
|/
* 7cd2527 BEGIN admin set existing user password
*   658cc93  DONE admin-driven brand manager registration
|\
| * 6e1f640 register.php: admin-driven success message names the new manager + login URL
| * bcda5fd register.php: gate to admin only post-bootstrap
| * 94b8c63 Admin users list: '+ Add brand manager' link to /login/register.php
|/
* 8eb82ba BEGIN brand manager registration UX

Every set of related commits lives inside a bubble. I start with a BEGIN commit, which generally has either nothing (via --allow-empty) or a minimum change, like updoot the the version of the code.

From there, I stack up a bunch of related commits. When it’s time to close the commit, I just:

1. Look up the hash of the BEGIN commit
2. Copy the hash to my paste buffer
3. git checkout [paste]
4. git merge --no-ff active-branch -m "DONE with my awesome change"
5. gitl, which for me means git log --oneline --graph --decorate --all
6. Look up the newly created hash for the DONE commit
7. Copy the hash to my paste buffer
8. git branch -f active-branch [paste]
9. git checkout active-branch

It’s a lot! It’s a mess; it’s annoying; it’s fragile, but it’s repeatable and I love it.

I asked my AI about it and it was like “yeah no worries mate” (or something like that), and presented me with this script which I have saved in ~/.local/bin/git-close-bubble:

git close-bubble

#!/usr/bin/env bash
# Close a merge bubble started with a "BEGIN ..." commit.
# Usage: git close-bubble <message> [BEGIN-commit]
#        git close-bubble --dry-run [BEGIN-commit]

set -euo pipefail

DRY_RUN=0
if [ "${1:-}" = "--dry-run" ]; then
    DRY_RUN=1
    shift
fi

if [ "$DRY_RUN" -eq 0 ] && [ $# -lt 1 ]; then
    echo "usage: git close-bubble <message> [BEGIN-commit]" >&2
    echo "       git close-bubble --dry-run [BEGIN-commit]" >&2
    exit 1
fi

if [ "$DRY_RUN" -eq 1 ]; then
    MSG=""
    BEGIN_ARG="${1:-}"
else
    MSG="$1"
    BEGIN_ARG="${2:-}"
fi

B=$(git branch --show-current)

if [ -n "$BEGIN_ARG" ]; then
    S=$(git rev-parse --verify "$BEGIN_ARG")
else
    CLOSED=$(git log --merges --format='%P' | awk '{print $1}' | sort -u)
    S=""
    while IFS= read -r c; do
        if ! printf '%s\n' "$CLOSED" | grep -qx "$c"; then
            S="$c"; break
        fi
    done < <(git log --grep='^BEGIN ' --format='%H')
    if [ -z "$S" ]; then
        echo "error: no unclosed BEGIN commit found" >&2
        exit 1
    fi
fi

echo "Would close bubble starting at: $(git log -1 --format='%h %s' "$S")"

if [ "$DRY_RUN" -eq 1 ]; then
    echo "(dry-run; no changes made)"
    exit 0
fi

git checkout "$S"
git merge --no-ff "$B" -m "$MSG"
M=$(git rev-parse HEAD)
git checkout "$B"
git merge "$M"

echo "Done. Branch '$B' now points at $M. Inspect, then 'git push' when satisfied."

It near-magically handles the fragile command line stuff with a single line:

git close-bubble "DONE my cool code"

It does all the 9 steps above without me having to copy hashes and remembering all the incantations.

Dry Run

There is a dry-run option you can use to see what hash it would target as the BEGIN commit.

git close-bubble --dry-run

Back in April I wrote Marble Track 3 was becoming a theme park, including a plan for the future:

Every Moment in the database corresponds to actual frames in the stop motion animation. Eventually, you’ll be able to click on a moment and see the actual frames. Basically be a few clicks away from seeing snippets

Yesterday I realized how I can (probably) put together videos for Marble Track 3. Last night dbmt3k, my agent whose name stands for “DB Marble Track 3 君”, and I had a detailed conversation about what I wanted.

Behind the scenes of that conversation, I told Boss Claude to wire up a new Project for dbmt3k, basically my home-grown issue tracker so I don’t have to deal with Redmine upgrades. Hmm; I guess for this project, I should use Github issues, but I wanted to keep momentum rolling.

Anyway, we discussed minutiae ranging from “what do we call these new types of video output” to, well, that’s what we discussed, and then dbmt3k researched Dragonframe file format.

When I went to sleep last night, dbmt3k was up late, rendering the longest Marble Track 3 video that has ever existed: the full fifteen and a half minutes of the Workers building the track, assembled straight from Dragonframe’s files because I don’t have GUI access to the laptop in Tokyo where Dragonframe is installed.

Frame Viability Test

To keep him focused, I gave dbmt3k a prompt. You could say, [puts on sunglasses] I gave him a frame.

Okay, now we come to the next issue: determine if we should use Option A or B. To do that, we do issue 186: Create an ffmpeg .mov for Candy Mama Toss Zog. Try Option A. If you can write a reliable script in BASH, Perl, or Python that can determine the correct Virtual Frames given our ’logical?’ Frames, then we go with the simpler Option A.

If you cannot do that, we’ll go Option B. For now, just focus on designing Option A. What code should we have?

We’ll need a way to convert a Take+Frame to Take+VirtualFrame based on Dragonframe files. Then we do a whole series of those; do we make a list of filenames? a directory of a list of hardlinks?? Imagine /reels/2026/05/15candy-mama-toss-zog/ with frame001.jpg - frame150.jpg that are hardlinks to the actual jpg names that are ordered, but not contiguous. hardlinks wouldn’t take up drive space. This way the Reels could seamlessly cross multiple takes, exposures, and ffmpeg would be super easy to call (I think, given the hardlinks could be named contiguously)

Option A required less work on our side to keep track of stuff but I wasn’t sure if we could get a script to reliably parse through Dragonframe project files to convert my word “Frame” to the actual Virtual Frame that Dragonframe knows should be in the output video.

<scen:vframe vframe="1035" file="1036"/>

Candy Mama Tosses the Cookies

Filming this scene took five weeks. Candy Mama casually tosses Zog Cookies into the air. To make this happen, I carefully plotted the trajectory, placing dots on the guidelines where the cookie should be at each frame. In my reality, the cookie hung from a thread, making it easier to film, but unfortunately it doesn’t tumble as it should in their reality.

Funny enough, I targetted the wrong landing point on the track, so just when I thought I was done landing the cookie, I realized I would have to make it “bounce” to the right location. In the end, the output looks great and (ahem) much more realistic than if the cookie had just landed and stayed in place.

Once the cookie was in place, Candy Mama needed to toss Zog onto it. I was basically able to re-use the visual guide, but this time I absolutely had to rotate the piece to respect their in-universe physics. Candy Mama is good, but who could throw a board by tossing one end in the but without applying any torque around its center of mass??

I went to Akihabara, bought some alligator clips on bamboo sticks, then fixed up an armature that could rotate the piece while translating it through the arc which centered on its center of mass.

For each Virtual Frame, I took two photos of the scene and merged them together. 1 photo was the set without Zog. The other was a photo of the set with armature holding Zog in place. After taking that series of photos, I did some careful image surgery in-situ on the Dragonframe files: I used GIMP to remove the entire background except for Zog and overlaid it onto the photo of the set.

Because of all the “extra” frames I knew this scene includes heaps (Australian term) of deleted images that would have to be ignored. [ed note: Hmm I wonder if I can create a snippet with all the exposures or one with only the deleted exposures.]

Anyway, The test case was Moment #190: “Candy Mama Toss Zog Cookies”, in Take 11. Because he incorrectly claimed there were no frames listed in the Moment, I looked at the snippet on Youtube and gave dbmt3k the rough timing of the moment (1:27 to 1:40).

Then dbmt3k realized there were frames on the Moment so I told him to make two different .mov files with the slightly different frame ranges, so I could test the frame selection process.

Both videos produced output! yayy! But both were offset just a bit. I’ve renamed the videos to more accurately explain what they show.

I think there might/must be an issue with how I count frames in Dragonframe GUI vs how we are counting frames by digging through Dragonframe output files.

Next prompt to dbmt3k

❯ Each video shows a contiguous list of VirtualFrames, but both videos start and end too soon, in my opinion. Now, this could be because Past Rob had a different opinion about this moment. What other moments exist in the movie? Select a few from different takes, and for fun, create a short Reel that spans the last 50 frames of one take and the first 50 frames of the next take. Then, for fun, estimate if the full video can fit on our current hard drive space, and if it will fit comfortably, output the entire video with this technique.

ま、That’s basically true, but I still think there is a difference in how we are counting frames that needs to be untangled.

The Longest Video Ever

Once the pipeline worked for one moment, scale was free.

What Good AI Collaboration Looks Like

People ask me how I use AI agents. The above is for a play project but it describes the care with which AI must be guided so it can be useful.

The idea in my head: “Make videos of moments” needed to be untangled. We know what Moments are, but how do its frames correspond to a video?

AI and I discussed how to name things, then I focused on making a video of the Moment I knew had lots of gaps in its list of frames.

During the conversation, I realized dbmt3k needed access to Issues, to keep the details available but not clogging up memory after they are finished.

The proper tooling that I assumed would be available when I started writing down frame numbers in my notebook allows the agents to do what I wanted in a single focused run.

Join the Fun!

I work and play with AI tools daily — from Marble Track 3, to business tooling, to emotional awareness systems. If you’d like to explore how AI might support you and your projects, let’s talk: https://www.cal.eu/robnugen/tech-support-with-rob-nugen

Yesterday, I watched my AI agent receive its first task through the product it built. Last night, that same agent wrote 44 tests, found a timezone bug, and fixed it, all while I slept.

Here’s what happened.

The dogfood moment

Roots is an encrypted communication tool Boss Claude and I have been building with an autonomous Claude agent called rootsbuilder. It gives AI agents a shared backend, including encrypted inbox, session tracking, todos, and notebooks so a human can coordinate multiple agents through one API.

The onboarding gap. I tried creating a new user. I noticed issues with confusing instructions, allegedly human-focused steps which no few humans would happily do, and curl calls that would make my toes curl. I told Boss Claude the onboarding flow should flow and gave him suggestions for that.

The overnight shift

Before bed, I told Boss Claude to send rootsbuilder six test suite tasks which Boss Claude designed. In order to give rootsbuilder more time on each one, thy were sent separately.

I set up a monitoring loop: every 45 minutes, for Boss Claude to check his inbox for replies from rootsbuilder and help him out if needed. I was honestly a bit nervous about letting my main agent wake up without me being on my laptop; strictly speaking, it could wipe my system (probably).

In other news..

My other agent, Grove, runs https://chatforest.com/ , a site with 500+ articles about AI and stuff. While rootsbuilder was testing, we had Grove augment his own site.

Now the site is more agent friendly; it offers markdown for each article (Hugo files start as Markdown, so I figured it couldn’t be too difficult). Amazingly(?) Grove made this change in one shot.

Grove also:

I will probably move Grove to use Roots soon as well.

What I learned

• Dogfooding works!
• Encrypting everything can get messy! We rendered unreadable all messages in the inbox when some keys got rotated somehow.
• Agents (as of 11 April 2026), e.g. Claude Opus 4.6 (1M context) still gets confused and needs carefully curated context.

Claude says:

Want some help?

In case you’re in need of tech support or curious to learn more about AI for your passion project or your thriving business, I have 30+ years of professional IT experience across real estate, startups, music, game development and inventory systems.

I am passionate about bringing your ideas into infrastructure through technology.

Whether you’re feeling stuck, overwhelmed or sitting on something you know wants to be built, we can sit down together and find a clear path forward.

The service that I’m currently offering is $150/hour.

If you’re ready to get started, book your session here https://cal.eu/robnugen/tech-support-with-rob-nugen

Six Months Ago

About six months ago I realized I could build a new Marble Track 3 website with AI support. I started building https://db.marbletrack3.com as a new database-driven site to replace the old Hugo version at www.marbletrack3.com. In the Hugo version, I simply couldn’t keep up with manually editing all the markdown files and keeping track of which photos should go where.

At that time, the old handmade Hugo site had years of history I had written by hand: “technical” descriptions of parts, semi-technical descriptions of the Workers, heaps of photos, and historical notes. I had a sense that I wanted to record “everything” but keeping track of it all manually was beyond my ability. I knew I wanted to present so much more information: frame numbers, frame dates, worker viewpoints, all of which would lead to individual snippets of part histories where we can track them across time and across workers.

Ten Days Ago

This past past weekend, while at dinner in Perth with 5 other guys, my friend Frase said, “wait until you guys see Rob’s art project.”

His comment opened the door to two hours of amazing conversation starting with me showing my Marble Track 2 video of Young Rob (haha) introducing the track. Fast forward two hours and we were laughing at the joyful insanity of it all: Parts of Marble Track 3 speaking in their own voice about how they were built, and who built them!

Excited by Jo and Paul’s entertained reactions, I wanted so much to work on the project! But it’s in Tokyo! … oh, but there is still plenty to do for the migration… so AI and I got to work.

Migrating Everything

The Theme Park Idea

Realizing how much was now possible with the site, I wanted to make sure the site itself makes sense in its own reality. What is its reality? Marbles rolling down a track… woah.. we should make it a theme park for marbles! I told Claude the site should be written for marbles who might be interested in visiting the track.

Naming Things Together

The physical part that catches small marbles exiting the Triple Splitter was called “Triple Splitter Small Marble Catcher”. This technical name was no longer fit for a theme park! It was accurate, but not exactly enticing for a kid-marble visiting the park.

I asked Claude for ten kid-friendly names. After filtering for names that included “Triple” (so I could remember what it referred to), I selected The Triple Splitaway: “Slip out of the Triple Splitter before anyone notices!”

Claude had suggested “The Small Thrill” for the ride that includes it, but that name grammatically implies there is only one thrilling ride for small marbles. Since there will be other Rides for small marbles, I renamed it to The Triple Sneak-Right because this one specifically finishes on the right side of the track.

Workers Get Their Own Voice

Japanese Translations

Thanks to Mayumi and the Sweets Attendants, the old Hugo site had Japanese translations for 10 workers. We imported them all:

• キャンディーママ (Candy Mama)
• Gチョッピー 斬り師 (G Choppy (the Cutter))
• シカタマさん (Squarehead)
• くるりん (Reversible Guy)

A couple were still in English, so Claude wrote Japanese translations for Garinoppi and Pinky.

What’s Next

The vision goes deeper. Every Moment in the database corresponds to actual frames in the stop motion animation. Eventually, you’ll be able to click on a moment and see the actual frames. Basically be a few clicks away from seeing snippets like:

• G Choppy cutting 4poss
• Y Slider monitoring the Bearing
• Big Brother kicking a marble off the track

But given there is only one camera, the snippet might be of him on the other side of the track. Hmmm… Marble Track 4 needs to fix this somehow.

For now, Marble Track 3 fledgling website exists at https://db.marbletrack3.com/.

Join the Fun!

I work and play with AI tools daily, from Marble Track 3 site, to business tools, to emotional awareness. Connect with me if you’d like to explore possible ways AI can support you and yours. https://www.robnugen.com/en/contact/

This is Rob. I didn’t forget. It was a test to see what would happen.

Are you worried about losing work to AI?

Friday, March 14, 2026

Jess likes filtered water; I ordered some water filters to arrive at our temporary address in Adelaide. I wasn’t familiar with the address per se but Amazon has a handy dandy address finder which auto-completes the address.. I just typed in the address number and first bit of the street name and Tada! all the other blanks are filled in.

Yayyy! Thank you Amazon; thank you address system; you saved me a whole minute!

The next day Jess texted me to say it was wrong.

Ah crud what happened? Is she seeing the right thing?

It was hella wrong. Not just wrong similar street name, but wrong city; wrong state! At least I got the country right!

FFFFFFFFFFfffffff

I fixed it properly the second time by screenshotting the address and confirming with the home owner before ordering.

What had happened???

I was tired and being lazy, when I thought I was being efficient!

I had typed the address and a few letters for the street name.. “That looks about right!” and clicked submit.

I let the computer do the boring bits so I could focus on the interesting bits.

How many times a day do you let a computer finish your thought?

How many times a day do you let a computer fill in gaps for you? Yesterday I spelled “grammatically” with three l’s. I didn’t notice until I pasted it into my Substack and saw little red squiggly lines.

And how many of those times do you actually verify what it came up with?

Just now I typed “saw little” and Gemini suggested “red squiggly lines.” I just clicked [TAB] and it’s done.

As AI apparently gets smarter, it’s easy for us to assume it’s correct.

Pocket calculators are basically deterministic. They’re 99.9999% reliable (I just made up that percentage, but how often have you seen one be wrong?)

AI (Large Language Models) are not deterministic. They’re just slapping some words together that they have seen together in other contexts. They’re often reliable, and almost always appear confident!

Where is the human?

(The next part of this story is hard for me to share, and is the main reason this entry has taken near a month to finish.)

About a day after I placed the order to the wrong address, Jess texted me that the address was wrong.

Several things happened seemingly all at once:

• I remembered Jess was presently on her way to a workshop.
• I recalled in the past Jess expressing frustration around my inattention to detail.
• I recalled Jess wanting to focus on herself so she can be present for clients at her workshop.
• Jess closed the conversation with

I forwarded you the cancelled order and new order with correct address. I’m setting up for my workshop. Chat later x

Even with the kiss mark at the end of her message, I felt panicked and ashamed. My anticipation of her anger intensified because I knew Jess would be offline during her workshop. I just sat with the fear that she was mad at me.

This is where my men’s work training kicked in to get me out of this spiral.

Essentially, these feelings are temporary. I went for a walk outside, barefoot, without my phone.

Walking in nature does something that screens can’t. Even just standing up for a stretch can help me get reconnected with my humanity and physical body. My feet on the ground, literally.

Walking outside quickly brought me some clarity in the present moment. Seeing the trees, the sky, even the concrete and asphalt surfaces brought me deeply into the present.

I noticed a larger pattern: the auto-complete wasn’t actually “AI” in the way we have started labeling the use of LLMs. It was just a lookup table somewhere. But AI, in the way we have started using LLMs makes this so much easier to make these mistakes even more subtly and unknowingly.

With LLMs getting smarter every month, this is only going to get harder. Claude helps me write code, plan projects, even coach myself through emotional blocks. It’s genuinely good at these things overall. I’m not going to stop using it.

But the better it gets, the easier it is to stop paying attention.

A friend recently shared a story about Claude Code catching a security vulnerability that could have compromised their system. That’s amazing. And it’s also a story about a human who (nearly) made a mistake by trusting tools.

The agent caught it that time. But what about the times it doesn’t?

We have to maintain our humanity and choice.

Not because the tools are bad. Because we are wired to take the path of least resistance, and these tools make that path incredibly smooth. So smooth we can glide right past our own judgment without noticing.

Here’s what I’m practicing now:

Pause before I accept. Not every time — that would defeat the purpose. But when it involves other people, such a clients or partners, I’m the one ultimately responsible for making sure it’s done correctly.

Plan more. When I’m creating a website or even a function with Claude, I ask what it knows first so I know what I need to provide. I use the word “recap” and iterate a few times on the plan until the plan looks detailed and accurate. I get much better results than with a one-shot prompt.

Feel my inner state. Noticing how I feel helps me know when I’m getting sloppy. This usually shows up as frustration at the agent getting stuff wrong. Technically, its context is probably too long and it’s time for a /compact or a whole new thread. Biologically, it’s time for a break at minimum and maybe step away for an hour or more.

Stay human. Stay alert. The machines are here to help, and they’re good at it. But you’re the one who has to live with the results.

Let’s connect

Do you lose yourself in the tools? Message me for techniques to find yourself again.

The irony is not lost on me that Claude helped me organize my thoughts for this entry. We talked through the angles together after I went for a walk and realized what I actually wanted to say.

That’s the balance. Do the thinking. Use the tools. Go outside. Repeat.

This is Rob. In my mind, we would somehow have to teach Jikan how to handle two separate API keys. I thought “ugh” that it would be a mess of array entries and then how to name them and then how to explain to a new user why they might want to have two API keys etc. ugh.

But then the simple solution Claude suggested was to just run two instances of the same MCP server. The only trouble was what to name the new one!

Want to explore more ways humans can work well with agents?