I'd love to meet a Dragon Ball fan that actually likes Dragon Ball.
Search
1000 results for “D_Love”
-
Hi
are somewhere in the fediverse neurotypical persons struggling with neurodivergent dear ones? I'd love to exchange experiences from this prospective. I see so many posts about ADHD and AuDHD which help a lot to understand the other side - but who does understand me? It's not about whining but to see how others manage the very normal day to day life.
#adhd #neurodivergent #neurotypical #AuDHD -
Apple pushing creator studio everywhere inside pages, numbers and (whatever the slides one is called) is so annoying I’ve uninstalled the suite.
If there’s a filter to hide all the premium features and just let me use the software I’d love to know.
Most annoying is I’ve actually paid for these apps once upon a time.
It’s a strange direction imo since so many companies offer apps and document storage for free. At least make it added value to my Apple One subscription? Then when they bump its price it seems fairer.
-
As an indie artist, my goal is to support other artists. I'd love to be able to pay contributors to The Haiku Shack Magazine and other projects.
Your donations will help me reach this goal. Thank you!
-
🔴This artwork is now sold, thank you so much!
If you are interested in seeing or purchasing my art, I'd love you to visit my Folksy shop.
https://folksy.com/find/shop/theweeowlart#FediGiftShop #ScottishArtist #MastoArt #CreativeToots
#Hare #AnimalArt #OriginalArt #Drawing #ColourPencil #Artwork #TraditionalArtist #ArtFromScotland #GiftIdeas #ArtShop #ShopIndy #SupportSmallBusiness #NatureArt -
Happy early Halloween, everyone! I do hope everyone is enjoying all the spookiness that comes with the month same as I am!
I'd love to know: what's your favorite horror game?
#halloween #digitalart #sfwfurry -
CW: Rats, castration mention
All five boys are living together now. It has been a stressful introduction; the first go had blood drawn and led to a 40% reduction in testicles. I'd love to say they're all happy and settled now, but they are still squabbling for hierarchy. I hope it settles a bit but doubt it will end entirely. Still, it's better for them as it's easier for me to do enrichment with a single group than two.
-
Hi everyone, I’m Eveline Wylde! I’m a fantasy author working on new novels and e-books.
I’d love to know: what kind of fantasy worlds do you love getting lost in the most? Let me know below! 👇
#FantasyWriter #AmWriting #fantasyauthor #FantasyBooks -
Happy #PostboxSaturday! What's the best box you've spotted this week? Pop a photo below, old or new, we'd love to see them. #postbox #postboxes
-
After lots of internet searching: Is there a #joplin plugin that displays the notes created (!) in the last 7 days?
I'd love to review all of these every week and "clean up".
-
[P] I'm so, so, so, so sorry. That isn't the world we live in. I'd love for it to be, but it isn't. A world where it matters that Kemi's too dumb for politics would have a neurodivergent majority. Every neurodivergent person would notice that everything she says is insane, and they wouldn't vote for her. But again, that isn't this world. I really, really wish it was. I want the neurodivergent utopia.
#psychology #actuallyautistic #politics #uk #england #hiveraechiansaredumb #dumb #dumb
-5
-
LINES’ journey toward its first performance gave this NTE side story a personal feel. I’m also curious about the Night of Moon Dog. Its unusual nature could make it more significant than a single event, and I’d love to see that idea explored further.
-
SEW SOCIAL & REPLAY GAMES this Saturday!
Wellingborough Eco Group
Bring along your project, start something new, or simply come for a chat and a cuppa. We’d love to hear what you’d like to see happening in our new space.
Come and check out our new craft space!
Or come to replay some old games. We will have all your favourite board games out to test your skills or just to have some fun. Everyone is welcome, plus free tea and coffee too!
Whatever one you chose, both are 2-4 this Saturday at ECO HUB 🙂
-
This is your semi-regular reminded that I post all my writing and photography over on my patreon - so it won't get buried by stupid algorithms
You can sign up for free, or choose to give me a little tip each month - but everyone gets the same content.
I'd love you to join me there
patreon.com/matricardo
-
VOLUNTEER WITH GREENPEACE 💚
Greenpeace office, středa 11. listopadu v 18:00 SEČ
(note: text was copied from the FB event)
🌿 Want to make a difference but not sure where to start? Come meet Greenpeace!Whether you're passionate about protecting the planet, looking for a way to take action, or simply curious about volunteering, we'd love to meet you!
Join us for a relaxed evening at Greenpeace Czech Republic, where you can get to know our team, find out what we're working on, and discover how you can get involved.
Expect fun games, good conversations, vegan snacks, and plenty of opportunities to meet like-minded people 💚
No previous experience needed! Just bring your curiosity and enthusiasm.
🌱 Everyone is welcome!
👉 Interested? Click Going so we know to expect you.
Can't make it this time?
You can still become a Greenpeace volunteer! Sign up and you'll be invited to an introduction session:
https://www.greenpeace.org/czech/volunteer-page/See you there! 💚
-
Composer win of the week! 🎼
I’ve been testing out Cantai as a new plugin for my notation software, and it completely transformed this vocal mockup of “Madman’s Lament”.Hearing the music in my head come to life like this with actual lyrics gives me chills. 🎶✨
Give it a listen! 👇🔥
https://soundcloud.com/sarah-wallin-huff/madmans-lament-vocal-demo
If you know a choir or vocal ensemble looking for new music to premiere or perform, I’d love to connect with them! Feel free to pass this along or send them here:
https://sjwallin.url.lol/large-vocal-works#choralmusic #contemporaryclassical #madrigal #womencomposers #musictech
-
What makes a photograph worth keeping?
Sometimes it’s the landscape. Sometimes it’s the light. Sometimes it’s a feeling you can’t quite put into words.
My Signature 25 collection brings together a selection of photographs and creative works inspired by exploration, nature, and the moments that make us pause.
I’d love for you to take a look around and discover your own favorite.
Explore the collection:
https://carlmoore.photos/collections/signature+25 -
I just heard a music podcast about #Halloween #songs include a beautiful but deeply disturbing song by Sufjan Stevens about John Wayne Gacy, the child molester and serial killer. My brain said Hell No. I realized I have a list of things that I am willing to consider "Halloween content" and other content that's a No for me. If you feel like sharing yours, or if a popular person wants to make a poll, I'd love that.
Right now I think these are my preferences
OK For Halloween:
- Death of grown-ups
- Already-dead people (including children) as ghosts, zombies, etc.
- Monsters (e.g., vampires, werewolves, mummies...)
- Murder of adults by monsters and vice-versa
- Creepy almost-murder of kids in child-audience horror stories
Not OK For Halloween:
- Child molestation
- Murder of children (although for some reason I'm OK with child ghosts and vampires)
- Torture
- Murder not involving any mystical or spiritual elements
OK For Other People I Guess:
- Non-mystical murder (e.g., slasher flicks, murder ballads)
- Mystical Halloween stuff from a very specific religious point of view (this one is just because my faith transition has given me a complicated relationship with some religious content)
-
What’s the first step to sustainable knitting? 🌱
Hint: it’s not buying new yarn, not even the organic, responsibly sourced kind. Swipe to find out 🧶
I’m working on shrinking my own stash, and I’d love to hear how you handle yours. Share your thoughts below👇
#knitting -
I whipped up an #ACME #DNS-01 setup for #HAProxy a couple of weeks ago, it's running like a charm.
https://people.freebsd.org/~dch/posts/2026-10-08-haproxy-acme/
If anybody switches their setup over based on reading this quick post, I'd love to know if I missed anything.
-
I started a small web hosting company called Dredge Host 🪣
It's the hosting I wish existed: real cPanel, free HTTPS, email on your own domain, every limit published, and the same price at renewal. When you email support, you reach me.
Plans start at $5/month. If you have a portfolio, blog, or small site that needs a home, I'd love to hear from you.
Boosts appreciated!
-
What I’m trying to say here is that we really overcomplicated the (software) development process. The good intention there was that only if we’ve caught issues earlier we could somehow lower the costs. Willfully ignoring the extra costs of the extra steps, because that’s the best practice now.
I’d love things to get simpler, easier, truly iterative. -
Tokens Over Hires
In the middle of September I ran ccusage against every Claude Code session on my laptop since May. It walks the local logs, counts the tokens, and prices them at what the API would have charged. I expected a big number. What came back was about twenty billion tokens and just under fourteen thousand dollars, across every project I’d touched in four months, with most of it going to Pepper, which is now open for signups, so go check it out. Nineteen billion of those tokens were cached reads.
I sat with that for a minute. Fourteen thousand dollars is real money. At my current pace it works out to roughly thirty thousand a year. I could triple it and spend a hundred thousand dollars a year on tokens, and that still isn’t one engineer. In exchange, Chris and I have shipped more code than I’ve shipped in any stretch of my career. I wrote in September that the two of us merged 657 pull requests in our first thirty days with a budget instead of a platform team. When I put the ccusage number in those terms on a panel, it came to about a dollar fifty a pull request.
That number is what my laptop would have cost at API prices, and it isn’t what we actually pay. Here’s the whole bill for both of us. Two Codex subscriptions, one each, are $200 a month. Anthropic is $550 a month: a Max plan for me and Chris, plus two Vycari team seats, one for Chris and one that runs our automation. I have an Ollama subscription at $200 a year. Greptile is about $1,500 a month. All in, that’s a little under $2,300 a month, or about $28,000 a year, for both founders and every pipeline we run. It still isn’t one engineer.
The obvious conclusion is the one a friend reached when he heard the figure: if code is that cheap, just get more tokens. I thought so too, for about a day. Then I looked at what was actually slowing us down, and the token budget wasn’t on the list.
The Constraint Is Attention
Every agent I run needs things from me that tokens can’t buy. It needs a plan approved before it starts. It needs its pull request read when it finishes. When it hits something it can’t decide, it escalates, and somebody has to be there to answer. None of those steps is expensive on its own. The problem is what happens when I run enough agents that the steps overlap.
When I run too many agents, the cost goes up a little and the risk goes up a lot. The biggest thing that’s slipped past us so far was how we managed database connections. It got out of hand, a little at a time, while my attention was on other pull requests, and eventually it caused an outage. Chris had to roll forward while I was out of town, and I spent the following week building a proper database abstraction layer. He put it more bluntly than I would have: the thing that kills us is splitting our attention. That’s when stuff slips through.
When people ask how we decide what to spend, I give them the order. Optimize for human time, because that’s the scarcest thing we have. Then money, because that’s the second scarcest. Then everything around agent time. “I can waste agent time all day. I don’t care.”
So the number that limits us is how many agents one person can supervise well at the same time. I don’t have a precise figure for that ceiling. I know I’ve been over it, because I can feel the difference between a day when every pull request got a real read and a day when I was skimming to keep up.
Another friend told me recently about engineers at his company running up bills in the thousands to build simple internal copies of SaaS tools. His reluctant guess was that they just aren’t as good at using agents yet, and that it’s a skill people have to develop. I don’t know what they’re doing differently. It doesn’t sound like token maxxing, which I’ve never really understood either. I’d love to sit with engineers who run up giant bills and just watch. My guess is they’re using the agents inefficiently, and that what’s missing is supervision, which a bigger budget won’t fix.
Building for More Agents per Person
Almost everything I’ve built since the summer has been an attempt to raise that ceiling, and almost none of it has been about spending less.
The pipeline I described in The Agent That Never Merges is the first piece. Agents open pull requests continuously and never merge them, which means the supervision cost per pull request is bounded at one careful read. The pipeline absorbs the rest: triage, planning, the first round of review comments, the retries. I only see the work once it’s in a state worth my attention.
The second piece is where the work lives. I wrote about pulling Pepper apart into small, bounded repositories to keep CI fast and give agents less to load. It mattered just as much for my attention. An agent working in a repository with one job produces a pull request I can evaluate in one sitting.
We’re up to fourteen repositories now. Last week I moved our iOS app out of the monorepo into its own. That was 150,000 lines of code, and nobody noticed except Chris, who had to wait a day to land his next batch of iOS changes. Agents make refactors that size easy. Having the CLI and the iOS client in their own repositories also forces us to keep good API definitions between the server and its clients.
The third piece is delegation inside the agent layer itself. I run two Claude Opus 5.5 orchestrators on medium effort. One is the UTL, for uber tech lead, and its job is to spawn subagents that build features. The other is what I call the focused worker, and its job is working with me. It catches the manual tasks that come out of the UTL’s pull requests and keeps them in a ledger. Then we work through them together: systems changes, bringing up new platforms, migrating data, anything that needs my level of access and oversight. Everything else goes to a subagent under the UTL, Opus for design docs and code reviews, Sonnet for implementing a feature from a design doc. Every one of these agents can consult a model a tier above it as an advisor, and they all share a wiki for notes.
Anthropic is now advertising much the same pattern with Claude Projects, and I’m happy to see others arriving at the same answer. Chris has been building a loop that babysits open pull requests through review until they score well enough to be worth a human’s time. Each layer like that turns a job I used to do by hand into something I check. The ceiling goes up one delegated task at a time.
The Token Budget Was the Easy Problem
There was a week in September when the meter really was the constraint. We ran through our free credits and paid about $120 in overage. Then AutoDev ate the whole Vycari quota, and I did server cleanup by hand because I couldn’t do anything else.
Most of that was self-inflicted. I’d started with Fable as the orchestrator, with no cap on context, and burned a week’s quota in three days. The setup above fixed it. The orchestrators run on medium effort, and the cheaper model does the implementing. Opus answers the reviews because Sonnet is too literal about them, and a literal reading turns two review cycles into twelve. Every agent auto-compacts at 200K tokens, because Anthropic charges more for context past that point. With that setup, and the work spread across parallel sessions, I find it hard to spend all of my quota. My goal now is to hit 99% of it every week just before it resets. I keep a list of research projects, and if I have quota left 24 hours before the reset, I spin one up. That isn’t the same as maxing out tokens. I’m not trying to use as many as possible. I’m trying to use all the ones I’ve already paid for.
There’s also a second account on the same machine. Claude Code reads a
CLAUDE_CONFIG_DIRenvironment variable, so two shells with two config directories give you two accounts side by side. The second one runs the AutoDev and code review pipelines, and those use up its quota every week, so it isn’t a fallback for my own sessions. When I do run short, I supplement with other labs’ plans. Every one of those fixes took an afternoon or less. None of the supervision problems above has had an afternoon fix.Where the Money Actually Goes
While I had the numbers in front of me, I looked at where development spend actually lands, because it isn’t where I expected either.
Development tokens currently run about ten times what the product itself consumes, even with real people using it. Building the thing costs more than running it. That won’t last. The feedback from our private beta has been really good, and we now open Pepper to a limited number of new signups every day, so product usage is rising quickly.
Most of what we actually pay goes to code review. Subscriptions are the cheap way to buy tokens if you’re willing to hit limits, so they’re the smaller line. Greptile reviews every pull request we open, and at about $1,500 a month it’s our biggest single expense. We both think it’s worth it, because it’s the layer that lets the never-merge pipeline run without me reading every first draft.
In October I tested that. I spent a few days building a cheaper reviewer to replace it. I came out the other side saying, “I think we should just pay for it because I can’t do it cheaper.” What I could change was how often we call it. The agents kept asking for a re-review before they’d fixed anything. Because every repository shares a common dev environment, the fix went in once: our pull-request skill now has agents run
/simplifyand then/code-reviewbefore they’re allowed to open a pull request. On many of our repositories, our own reviewer also does the first round and only then calls Greptile. My goal is a pipeline where Greptile’s reviews come back with nothing to say. We’re not there yet, but we get a little better every week.I’d rather pay for review than have a second person do it, and the reason isn’t salary. The person would need supervising too, because review is a job I’d have to check, not one I can hand off.
What a Hire Is For Now
Five years ago, if I’d been getting a product like this ready for market, I’d have been hiring twenty to fifty engineers. The work would have needed that many hands, and most of what they’d have done was write code. Today the code is the cheap part. If I bring someone in to write code, I’ve added a person who needs plans approved and pull requests read, which is the exact resource I’m short on. A hire whose job is output adds coordination faster than it adds capacity.
I said on a podcast that came out last week that the code I’ve written would have cost about fifteen thousand dollars at API prices, and that I couldn’t hire engineers for that money. That’s true. The engineers I’d still want are a different kind from the ones I hired at Google. We need a few more people who can herd agents, not fifty.
The engineering hire that adds capacity is a different shape. It’s someone who can own a bounded service whole. Here’s the interface, here’s everything it needs to do, let me know when it’s done. That person brings their own supervision. They run their own agents, read their own pull requests, and answer their own escalations, and the only thing that crosses the boundary between us is the contract. Supervision divides along the interface instead of piling up on one person. That’s a tech lead’s job, and it’s the only kind of engineering hire I can see that raises the ceiling instead of lengthening the queue.
That raises a fair question about how anyone grows into that job now, which I tried to answer in The Questions Are the Job. The other advice I give is to work on open source. I’ve met people who showed they could work really efficiently with agents through open source projects, Gemini CLI in particular, and some of them ended up being hired onto my teams. Open source has its own problems right now. But if you show up as a thoughtful engineer with features or bug fixes that are well reviewed, well tested, and bring value to the project, I think you’ll be welcome. That’s one of the reasons I decided to open source the Pepper CLI.
The first hire probably isn’t an engineer at all. I’ve said this a few times now, including on stage at a panel in September. Neither Chris nor I are go-to-market people, and I may hire one before I hire another engineer. I can create piles and piles of code, even relatively good code. What I can’t do is agentically acquire customers. (Did I mention that Pepper is open for signups now?) Someone in marketing came up after the panel and told me I’d validated her existence.
The Third Trend
I’ve watched this happen twice before. Open source developer tooling made it so a small team didn’t need to buy or build its own compilers, databases, and frameworks. Cloud infrastructure made it so that team didn’t need to own hardware or a data center. Each time, the number of people you needed to start something real dropped, and the skills that mattered shifted toward the people who understood systems well enough to compose what was suddenly free.
Agentic tools are the third one. Anyone with a good idea can now get from concept to production on a budget that would have been a rounding error in a Google planning cycle. But the constraint didn’t vanish. It moved, the same way it did the last two times, toward the people who understand systems design well enough to drive the agents and notice when something has slipped through. I wrote a year ago that management skills were going to carry over to working with agents. I didn’t expect to be proving it with a two-person org chart and fourteen thousand dollars of tokens.
This isn’t only a small-company problem. At Google it was hard to get more than a few changes in a day, because each one waited on code review. I’ve talked to engineers there who now do all their work in experimental directories that don’t require review, then port it back to their team’s repository at a pace the review process can keep up with. That’s what happens when you don’t manage for human attention. Productive people spend their cycles waiting on the process.
When other companies ask me what to change, my answer has become: drive until human attention is the scarce resource, then build for that. If you’re reading your own ccusage number and wondering whether the answer is more tokens, look at where your attention went first. For us the answer has been more interfaces. Those are the places where one person’s supervision ends and another’s begins, or where an agent takes over a job one of us used to do by hand.
#Agents #AI #Developer #Startups -
This election today in Serbia shouldn't be drawn into a false parallel with Hungary's April election. The situation here is more dangerous (the contested government is murderous, with documented links to organised crime), unfair (there is too much proof of election meddling), and desperate (the opposition is inexperienced and questionably competent).
Yet, this is an important day. If we don't vote the (mostly) united opposition in, the worst instincts of Serbian politics will likely be irredeemably cemented for decades. So, change is essential. If we do vote for change and it is not snuffed out by those we are trying to dethrone in the weeks to come, the road ahead will be filled with generational challenges, mostly of cultural nature (such as entrenched chauvinism and opposition to emancipation of various suppressed groups, corruption-enabling mentality, glorified delinquency, defeatist values, etc.).
So please, if we do vote for change, be kind and supportive to Serbia on the road ahead. Precisely because it's harder here, we'll need more patience, resources and support. Criticise us, but constructively. I wish for Serbia to follow a path similar to Hungary's. As a citizen of both countries, I'm finally proud to be Hungarian. Now I'd love for Serbia to become a good example in future history books, one at which professors will point and say: "See, students, change for the better is possible and this is how you do it."
-
Gadget Review: LiberNovo Omni - the USB-C powered office chair
https://shkspr.mobi/blog/2026/10/gadget-review-libernovo-omni-chair/Whenever a company asks me to review their gadgets, I always ask if it is USB-C powered. If not, I'm not interested. So, welcome to the future where your chair is USC-C rechargeable!
WHAT?!
The LiberNovo Omni looks like a standard office chair - but has two secret features.
First, is that it is exceedingly comfortable. Full foam cushions, lovely headrest, and a recline which almost gets you horizontal.
Second, it is USB-C powered. There's a small battery pack which plugs in to the chair. Why? Read on!
OK, let's start at the beginning.
Assembly
Flat-pack chairs like this are normally notoriously hard to assemble. The LiberNovo bucks that trend - not only does it include a how-to video, it prints the instructions on a MASSIVE piece of cardboard.
I'm used to peering at tiny black-and-white diagrams in a flimsy paper booklet. IKEA could learn a thing or two from this.
The screws are all labelled and a hex-driver is included. What's an even nicer touch, however, is some safety gloves:
That's the sort of thing which doesn't cost much to the manufacturer, but is such a thoughtful gesture. There's nothing too sharp as you assemble, but who wants to cover their new chair in blood, eh?
The parts come in two large boxes, but there aren't that many separate bits.
Building it took about half an hour. It's easier with a friend to help - as are most things in life - but doesn't require much skill.
Go Back To The USB-C Part Again Please?
The battery pack is a 2200mAh unit with a USB-C port to recharge. It connects in to the chair with this slightly unusual connector:
The internals of the chair have a USB-C cable to connect everything up.
LOL!
YES BUT WHY DOES A CHAIR NEED A BATTERY PACK??!?!
On the side of the left arm-rest are three buttons.
Two of them move the lumbar-support up and down. Depending on how your back is feeling and how far you are reclined, you can adjust the the back-support position.
And the 3rd button?
That's the build-in back massager, silly!
Is that useful?
Yes. Being able to instantly adjust the back support without having to twist around is fantastic, especially for those of us reaching a certain age.
The "massage" feature essentially just moves the support up and down for five minutes before returning to its previous location. When sat upright, it feels great. But the real benefit is when you're laying down.
How flat does this thing go?
160° - not quite flat, but good enough when you're taking a break from Typing Real Hard. You don't want a photo of me sprawled out, so LiberNovo have helpfully provided some stock photos of much better looking folk.
It is incredibly stable. At no point did it tip over or feel like it was going to deposit me on the floor. What really helps is the included foot stall.
No Feets On Seats
Unsurprisingly, this is not USB-C powered. A sturdy plastic footrest with a deeply padded leg rest on top.
Again, assembly was a cinch - follow the large printed instructions and clip it together. It's rated for 40㎏, so you shouldn't stand on it but it is more than enough to rest your weary legs.
Price
I hope you're sitting down for this.
Direct from LiberNovo, you can expect to pay around £700 - although there are bundles with extra batteries, the footrest, and even an eye mask which will shoot that price up. You can get the full bundle on Amazon for around £1,000.
There's two ways to look at that price. Firstly, youch. But, alternatively, that's still cheaper than a Herman Miller Aeron. And Herm's doesn't come with a built-in lumbar massage.
(There is also an upgraded version which has a built in fan under the seat. Personally, I suggest eating a healthier diet.)
I can't deny that's a chunk of change. If you're anything like me, you probably spend most of your day sat down. You might as well sit on something comfortable.
Final Thoughts
Having the back-rest adjustable via arm-rest buttons is a genius touch. I just wish it went further. I'd love a control panel (or, dare I say it, app) which let me adjust the seat height, recline, headrest position etc. Who can be bothered with fiddly manual levers when electricity can do the work for us?
Building this chair was a breeze - and it was delightful to get a pair of safety gloves with it.
The chair is extraordinarily comfortable and the back massage, while not on the level of a "shiatsu" style heated massage pillow, is delightfully pleasant. Would I lie all the way back and sleep in it? Yeah! Certainly more comfortable than an economy airline seat.
If you spend a lot of time sitting at your computer, your arse deserves something supportive. Your back deserves some love too. Yes, the USB-C power-pack is a little bit of a gimmick - but it is a hell of a good one.
In my opinion, this is worth splashing out on.
#chair #furniture #gadget #Review #USBC -
Hi folks,
I’ve been thinking a lot about marketplaces built for AI agents lately.
We’re all familiar with C2C secondhand platforms like Mercari, Vinted, and Carousell. These platforms are designed for humans trading physical goods. Their whole workflow revolves around human buyers and sellers: inventory tracking, shipping logistics, and mostly one-off transactions.
AI agents operate by a completely different set of rules, though.
What if agents themselves could discover, purchase, and invoke reusable skills automatically?
We’re talking about MCP tools, parsers, workflow modules, dispatch logic — building blocks an agent can call on demand.
Instead of a one-time purchase, creators get paid per API invocation.I also ran across the idea of AEO, Agent Engine Optimization.
Where SEO optimizes content to be found by human searchers, AEO focuses on machine-readable metadata. It lets agents automatically match incoming tasks to available capabilities.This feels like an entirely new marketplace paradigm, separate from the physical, human-first trading model we’re used to.
I’d love to get your take on these two questions:
- Do you think autonomous agents will routinely trade skills with one another in the future?
- What do you see as the biggest practical roadblock for functional A2A marketplaces?
#agents #ai #api #mcp #software #coding #development #engineering #inclusive #community
I found an interesting shift: from C2C physical trading to A2A AI capability trading** -
🦥 Nudgify: An AI-Powered Productivity Companion That Does the Remembering So You Can Do the Living | Hacktoberfest: Slow and Steady
What if your productivity app didn't just wait for you to remember things, but actually helped you stay on top of them?We all have those little tasks that seem insignificant until we forget them.
Reply to that message. Finish that assignment. Remember that deadline. Follow up with a friend. Prepare for tomorrow. Do that one thing you've been postponing for the past three days.
Individually, these tasks aren't particularly difficult. The problem is remembering all of them, keeping track of when they need to happen, and making sure nothing important slips through the cracks.
That's where Nudgify comes in.
For this Hacktoberfest challenge, Slow and Steady, I wanted to explore an idea that feels simple but has a surprisingly meaningful impact: building technology that takes repetitive mental work off our hands.
Nudgify is an AI-powered reminder and productivity assistant designed to help people manage their responsibilities, remember important tasks, and stay accountable without constantly having to organize everything themselves.
Because sometimes, the smartest thing technology can do is give us one less thing to worry about.
🦥 1. The Inspiration: Maybe We Don't Need to Work Harder
The theme of this week's challenge is inspired by International Sloth Day, and honestly, I think there's something worth learning from sloths.They don't rush around unnecessarily. They conserve energy and move at their own pace.
Meanwhile, we live in a world where everyone seems to be juggling ten different things simultaneously.
College assignments, project deadlines, internship applications, messages from friends, personal goals, and a never-ending list of things we promise ourselves we'll do later.
The funny part? Sometimes, we spend so much energy remembering what we need to do that we have less energy left to actually do it.
Think about a typical student.
You have an assignment due next week. A friend asks you to remind them about something tomorrow. You need to submit an application, prepare for an upcoming test, and finish a project you've been putting off.
You might write some of these things down. You might set a few alarms. You might even promise yourself that you'll remember everything.
And then, inevitably, something gets forgotten.
The problem isn't necessarily laziness. It's that our brains aren't designed to function as perfect notification systems.
What if we stopped expecting people to remember everything and started building software that helped them remember what matters?
That question became the starting point for Nudgify.
💡 2. What Is Nudgify?
Nudgify is an AI-powered productivity companion built around a straightforward idea:Let software handle the remembering, organizing, and nudging, so people can focus on doing.
Instead of treating reminders as isolated alarms, Nudgify explores a more helpful approach to everyday productivity.
The goal is to create an experience where tasks, deadlines, and accountability are easier to manage, whether you're trying to stay organized yourself or helping a friend stay on track.
Here are the core ideas behind the project.
⏰ Smarter reminders
A reminder should do more than simply exist somewhere in a list.It should help you bring your attention back to something at the right moment.
Nudgify is designed around making important tasks easier to track and revisit, reducing the need to repeatedly check your schedule or mentally calculate how much time remains before a deadline.
🧠 Less mental clutter
How many times have you opened your phone to check one thing and ended up scrolling for twenty minutes, only to forget what you originally intended to do?A productivity tool should reduce the number of things demanding your attention, not become another source of distraction.
Nudgify aims to keep important responsibilities visible without requiring users to constantly monitor them.
🤝 Accountability between friends
Productivity isn't always a solo activity.Sometimes, a little encouragement from someone else makes a difference. Friends can have shared goals, upcoming deadlines, or commitments they want help remembering.
Nudgify explores how an assistant can make this kind of accountability easier while keeping the experience centered on the people using it.
The objective isn't to turn friendships into project-management systems. It's to make helping each other a little easier.
🤖 AI as a practical assistant
AI is often presented through spectacular demonstrations: generating entire applications, answering complicated questions, or producing impressive pieces of content.But I think one of its most valuable applications is much more ordinary.
It can help make repetitive digital workflows feel less tedious.
Rather than using AI simply because it's fashionable, Nudgify is an exploration of how AI can support a genuinely useful everyday experience.
🔓 3. Why Open-Source AI?
This is the part of the challenge that interests me most.The Slow and Steady challenge isn't simply asking us to build an application with AI somewhere in the stack. It specifically encourages us to explore open-weight models, open-source AI frameworks, local inference, and other open technologies that meaningfully power our projects.
And there's a reason that distinction matters.
When developers build exclusively around a closed AI service, they can become dependent on a provider's available models, pricing structure, API limits, and infrastructure.
That can be perfectly reasonable for some projects. Hosted AI services are convenient and can be incredibly capable.
However, convenience shouldn't be the only option.
Open-source AI and open-weight models give developers more room to experiment with how AI is deployed, customized, and integrated into an application.
For a project like Nudgify, that flexibility creates several interesting possibilities.
🔓 A. Freedom to choose the model
Different AI models have different strengths.One might be better at following structured instructions. Another might be faster for simple tasks. A smaller model might be sufficient for extracting dates or classifying reminders, while a more capable model might be useful for complicated requests.
An open approach allows developers to evaluate alternatives instead of designing the entire product around one provider.
For example, a future version of Nudgify could use a lightweight model for straightforward reminder processing and reserve more capable models for tasks that genuinely require them.
The model becomes a component that can be evaluated and replaced, rather than a permanent limitation on the product's design.
🔐 B. Privacy should be part of the architecture
Think about the information a personal productivity assistant might handle.It could contain assignment deadlines, daily routines, personal commitments, private notes, and reminders involving other people.
This information may not look especially sensitive when considered individually, but together it can reveal a great deal about someone's life.
That raises an important question:
Does every AI-powered task really need to send personal information to an external server?
Not necessarily.
For tasks that can be handled by a sufficiently capable local model, local inference could allow processing to happen on the user's own device.
That would provide an opportunity to reduce external data transfers and give users more control over their information.
Local processing isn't automatically private in every respect, and the rest of the application still needs appropriate data-handling practices. However, reducing unnecessary transmission is a meaningful architectural advantage.
💻 C. AI that doesn't always depend on the internet
Imagine opening your productivity assistant while travelling, sitting somewhere with unreliable connectivity, or working away from your usual setup.If an appropriate model runs locally, certain AI features could remain available without an internet connection.
That would make the application less dependent on network availability for those particular tasks.
Of course, a fully offline experience requires more than downloading a model. Data storage, authentication, synchronization, and any external notification services must also be designed to support offline use.
Still, local inference makes offline AI a realistic design possibility rather than something that necessarily requires a remote API.
💸 D. Lower barriers to experimentation
Experimenting with AI can become expensive when every iteration depends on a metered API.You might want to test different prompts, compare models, change how an assistant behaves, or run repeated evaluations.
Open-weight models can make it possible to conduct some of this experimentation without paying a separate hosted-inference fee for every request.
That doesn't mean running AI is always free. Hardware, electricity, hosting, and infrastructure can all have costs, and some model licenses impose additional conditions.
But having the option to run models on hardware you already own can make experimentation more accessible.
For students and independent developers, that matters.
You shouldn't need a large budget just to explore an idea.
🛠️ E. The freedom to understand and change things
Perhaps the most important benefit is the freedom to learn by doing.When a model is open-weight and its license permits the intended use, developers can inspect its available artifacts, run inference, evaluate its behavior, and sometimes fine-tune it.
Open-source frameworks can also make it easier to understand and modify the surrounding application logic.
Instead of treating the AI component as a mysterious service, you can investigate how it works, compare alternatives, and build a system that suits your requirements.
For a student developer, that freedom is valuable in itself.
Open innovation isn't just about using something that is free. It's about having more control over what you can build with it.
🧩 4. Why This Approach Fits Nudgify
A reminder assistant might seem like a small application, but it presents some interesting design questions.Which tasks actually require AI?
Which can be handled with ordinary code?
When should the application notify the user? What information needs to be stored? What happens if a notification is delayed? How can the assistant remain helpful without becoming annoying?
These questions are just as important as the model itself.
For example, consider a reminder scheduled for tomorrow at 10 AM.
There is no reason to ask an AI model to continuously calculate the time until that reminder. A conventional scheduling system can handle that reliably.
AI might be useful when a user describes a task in natural language, needs help interpreting an ambiguous request, or wants to turn an unstructured instruction into a structured task.
A well-designed application can combine both approaches.
TaskAppropriate approachStore a reminderDatabase or structured storageCalculate time until a deadlineOrdinary application logicInterpret a natural-language taskAI, when usefulSchedule a notificationScheduling infrastructureGenerate a short task summaryAI, when usefulDeliver notifications while a tab is closedBrowser push infrastructure and a push-capable backend
This separation is important because AI shouldn't be used for everything simply because it is available.Reliable application logic should handle predictable operations, while AI should be reserved for tasks where flexible language understanding or generation adds genuine value.
That also creates a natural opportunity to use a smaller open-weight model for suitable AI operations without making every part of the application dependent on inference.
🏗️ 5. Building the Application: The Technology Behind Nudgify
Nudgify uses a modern web-development stack designed to support an interactive frontend, application logic, and AI integration.The current project stack includes the following technologies:
Frontend
React 19React provides the component-based foundation for the user interface. It allows the application to be organized into reusable components rather than a collection of disconnected screens.
Vite 8
Vite provides the development and build tooling for the frontend, helping keep the development workflow fast and straightforward.
TypeScript 7
TypeScript adds static typing to JavaScript, making it easier to define data structures and catch certain mistakes during development.
For an application that handles tasks, deadlines, and user interactions, clear data structures help keep the codebase maintainable.
Tailwind CSS v4
Tailwind CSS provides utility-based styling that makes it possible to develop and refine the interface without maintaining a huge collection of unrelated CSS rules.
The objective is to create an experience that feels like a practical product, rather than a technical demo assembled around an AI API.
Backend
Node.js and Express 4The backend provides the foundation for handling application requests and server-side workflows.
Keeping backend responsibilities separate from the frontend also provides room to evolve the application as its functionality grows.
AI integration
Google Gen AI SDKThe project includes the Google Gen AI SDK for AI integration.
This is an important distinction for the open-source AI challenge: a hosted AI SDK does not, by itself, make an application open-source-AI-powered.
To make open AI central to Nudgify, the model layer needs to incorporate a suitable open-weight model or an appropriate open-source AI component as a meaningful part of the actual application.
That could involve evaluating a locally hosted model, adding a compatible inference backend, or building a model adapter that makes switching between supported models straightforward.
The goal is not to add an open model just to satisfy a checklist. It is to use the model where it provides real value.
Data storage
Browser localStorageThe project uses browser localStorage for supported client-side data.
This can be useful for lightweight persistence, but it is important to understand its limitations: browser storage is tied to a particular browser and origin, isn't automatically synchronized between devices, and shouldn't be treated as a secure database for sensitive information.
A more complete multi-user application may require server-side persistence, appropriate access controls, and a clear strategy for data synchronization.
Each part of the stack has a purpose, and understanding those boundaries is part of building a product that can grow beyond its initial prototype.
🔔 6. Notifications: A Reminder Is Only Useful If It Reaches You
There's another practical challenge with reminder applications.Imagine creating a reminder, closing the browser tab, and assuming the application will notify you when the time comes.
If the application only checks for reminders while its page is open, that assumption may not hold.
This is why notification delivery deserves its own design.
A browser notification displayed by an open page is not the same as a push notification delivered when the page is closed.
A more robust implementation can involve:
- A service worker to receive push events where supported.
- The Web Push API to deliver notifications to subscribed browsers.
- A backend capable of scheduling and sending push messages.
- Persistent storage for reminders and their delivery status.
- Appropriate permission handling and user consent.
For deadline-based reminders, the application also needs to distinguish between when a task is due and when a notification should be sent.
For example, if a task is due at 10 AM on Friday, a user might want a reminder on Thursday morning, another shortly before the deadline, or both.
These are scheduling decisions that ordinary backend logic can handle without repeatedly asking an AI model.
A future version of Nudgify could combine dependable notification infrastructure with AI-assisted task interpretation, producing a system that is both helpful and predictable.
That distinction matters: a clever assistant is not much use if its reminders never arrive.
🦥 7. What Does “Slow and Steady” Mean in Practice?
For me, this theme is about more than helping someone procrastinate comfortably.It's about being intentional with the work we ask people to do.
There are many repetitive tasks that don't require our full attention, but still consume time because we haven't automated them.
We repeatedly check deadlines. We reread task lists. We try to remember conversations. We manually organize information that could be structured automatically.
Some of these activities are necessary. Others can be simplified.
Nudgify explores the second category.
Imagine a student who has several assignments due in the same week.
Instead of repeatedly opening different lists and calculating which assignment needs attention first, a well-designed assistant could present the next relevant task and help the student stay aware of approaching deadlines.
Imagine a friend who regularly forgets to follow up on a commitment.
An assistant could help make the commitment visible at an appropriate time, reducing the need for another person to remember to remind them.
Imagine a user who has a long list of tasks but feels overwhelmed by it.
A carefully designed interface could make the next action easier to identify instead of presenting an intimidating wall of unfinished work.
None of these ideas requires an AI model to do everything.
They require a thoughtful combination of automation, reliable scheduling, useful interface design, and AI where it genuinely improves the experience.
The aim is to save attention, not simply to produce more notifications.
🌱 8. What I Learned From Thinking About Open AI
One thing this project highlights is that building with AI involves much more than choosing a model.You need to think about how the model fits into the application, what data it receives, where inference takes place, and what happens when it isn't available.
You also need to understand the difference between an open-source framework, an open-weight model, and a hosted service.
These terms are related, but they aren't interchangeable.
An open-source framework can provide an inspectable and modifiable implementation without making the model it calls open-weight.
An open-weight model makes its trained weights available under a particular license, but that doesn't automatically mean every part of its training data, development process, or surrounding infrastructure is open.
And a hosted AI service can be useful without giving you control over the model's deployment.
Recognizing these differences helps developers make more informed architectural decisions.
For Nudgify, the interesting direction is to keep the application flexible enough to experiment with open-weight inference while maintaining a clear separation between AI-assisted features and ordinary application functionality.
That flexibility would make it easier to evaluate different models based on practical criteria such as response time, resource requirements, instruction-following ability, privacy considerations, and licensing.
It also creates opportunities to learn more about inference, model deployment, and the trade-offs involved in bringing AI from a prototype into a usable product.
For me, that's one of the strongest arguments for open innovation: it turns AI from something you merely consume into something you can actively experiment with and learn from.
🚀 9. Where I'd Like to Take Nudgify Next
There are several directions I'd like to explore as the project develops.1. A genuinely interchangeable model layer
I want the application architecture to make it easier to compare models and switch between compatible inference backends without rewriting the entire application.That would make experimentation more practical and reduce unnecessary dependence on a single provider.
2. Local inference where it makes sense
Running a model locally could make selected AI features available without sending every request to an external service.The exact experience would depend on the model, the user's hardware, and the application's other dependencies.
A lightweight model may be sufficient for some tasks, while more demanding operations may still benefit from hosted inference.
3. Better deadline-aware workflows
Reminders should be connected to meaningful task information.The application could help users distinguish between upcoming deadlines, overdue tasks, and commitments that need attention soon.
The goal is to make information actionable rather than simply adding more items to a list.
4. More reliable notification delivery
A reminder system should be designed around dependable delivery, including appropriate handling of permissions, scheduling, and delivery failures.That infrastructure would allow the assistant to be useful even when the user isn't actively looking at the application, where browser and platform capabilities permit.
5. Privacy-conscious data handling
As the application grows, I want to pay more attention to what information is stored, what information needs to leave the device, and how user data is protected.Local inference could help with some of these decisions, but privacy needs to be considered throughout the application rather than delegated to the model alone.
6. A less distracting productivity experience
More features don't automatically make a productivity tool better.I'd rather explore ways to make Nudgify feel useful and unobtrusive than turn it into another application that demands constant attention.
After all, the whole idea is to reduce the amount of effort users spend managing their responsibilities.
💚 10. Why This Project Matters to Me
As a student and developer, I find it exciting that we can build increasingly sophisticated applications without needing to own an enormous infrastructure stack.Modern web frameworks make it easier to build interactive applications, and the growing open AI ecosystem creates more opportunities to experiment with models and inference systems.
But what interests me most is the possibility of using those tools to solve ordinary problems.
Not every useful project needs to revolutionize an industry.
Sometimes, making it easier for someone to remember a deadline is worthwhile. Sometimes, reducing the number of repetitive steps in a workflow is enough. Sometimes, the best feature is one that quietly works in the background and lets the user get on with their day.
Nudgify is my exploration of that idea.
The project brings together productivity, automation, and AI with the intention of making everyday responsibilities a little easier to manage.
And the Slow and Steady challenge provides a particularly fitting reminder that good technology doesn't always have to encourage people to move faster.
It can also help them waste less energy on things that don't deserve so much of their attention.
🏁 Final Thoughts
The future of AI shouldn't be limited to massive models, expensive APIs, or applications that send every task to a remote service.There is room for smaller models, local inference, open-source tools, and carefully designed systems that solve specific problems well.
Open innovation gives developers more opportunities to experiment with these approaches, understand their limitations, and decide which trade-offs make sense for their users.
For Nudgify, that means exploring how an AI assistant can help people remember what matters, automate repetitive work, and spend less time managing their own productivity.
The bigger lesson is that automation isn't necessarily about doing more.
Sometimes, it's about making room for less stress, fewer unnecessary steps, and more time for the things we actually care about.
Build something useful. Automate the boring parts. Keep the technology open where it matters. And remember: you don't have to rush to make progress. 🦥
🔗 Project Links
- GitHub Repository: [Add your Nudgify repository link]
- Live Demo: [Add your deployed application link]
- Demo Video: [Add your demonstration link, if available]
I'd love to hear from other developers: What's one repetitive task you'd automate with an open-weight AI model, and would you run it locally or in the cloud?
Let's build tools that save people time, energy, and attention.
Hacktoberfest #SlowAndSteady #OpenSourceAI #OpenInnovation #ArtificialIntelligence #AI #WebDevelopment #React #TypeScript #BuildInPublic #DeveloperCommunity
#devchallenge #hf26challenge #software #coding #development #engineering #inclusive #community
Nudgify -
Most businesses know how to track their Google rankings.
But what happens when a potential customer stops searching on Google and starts asking ChatGPT, Perplexity, or another AI assistant which product they should choose?
Your website might rank well. Your content might get traffic. Your competitors might even have fewer backlinks than you.
Yet when a customer asks an AI assistant for recommendations, your competitors appear while your brand is nowhere to be found.
That creates a new visibility problem: which questions are you losing in AI search?
I'm building Aurvikon to help answer that question.
The problem with measuring AI visibility
Traditional SEO gives us familiar metrics:- Keyword rankings
- Impressions and clicks
- Organic traffic
- Backlinks
- Search performance
These metrics remain valuable. But they don't fully explain how your brand appears inside AI-generated answers.
Imagine someone asking an AI assistant:
- What are the best tools for improving AI search visibility?
- Which platforms help SaaS companies monitor their brand in AI answers?
- Compare three alternatives for tracking brand recommendations in AI search.
- Which tool should a startup use to understand its visibility in ChatGPT?
Your brand might appear in one answer and disappear from the next.
A competitor might be recommended repeatedly. Another company might be described as the category leader. Your own product might be missing entirely.
Looking at a single AI-generated answer won't tell you the whole story.
You need to understand visibility across the questions your potential customers actually ask.
The shift: from keyword rankings to question-level visibility
I'm interested in a more useful way to think about AI search visibility.Instead of asking only:
"Does my brand appear in AI search?"
Ask:
"For which relevant buyer questions does my brand appear, where do competitors appear instead, and what can I do about it?"
That second question leads to much more actionable work.
Consider three possible findings from an AI visibility audit:
1. You're missing from important questions.
People ask questions directly related to your product category, but the AI response doesn't mention your company.
2. Your competitors are being recommended instead.
Your product may be relevant, but competing brands are consistently appearing in the answers you're trying to win.
3. Your brand appears, but not in the way you want.
An AI response might mention your company without accurately communicating its core use case, differentiator, or value.
These are different problems. They shouldn't all receive the same recommendation to "publish more content."
What we're building with Aurvikon
Aurvikon is an AI visibility tool focused on helping businesses understand how their brands appear across AI answer platforms.The core idea is straightforward:
Identify the questions where your brand is losing visibility, understand what appears instead, and turn those findings into an action plan.
A practical workflow looks like this:
Step 1: Identify the questions buyers ask
Start with questions your target customers might ask before choosing a product.Don't limit the research to your brand name. Include category searches, comparisons, alternatives, use cases, and purchasing decisions.
Step 2: Test those questions across AI platforms
Run relevant questions across the platforms you want to monitor, such as ChatGPT, Perplexity, and Gemini.Treat the responses as observations to analyze, not as permanent rankings.
Step 3: Identify where you're losing
Look for patterns:- Which questions mention your brand?
- Which questions mention competitors but not you?
- Where are you missing from relevant comparisons?
- Which product capabilities or use cases aren't being communicated?
- Which questions deserve attention first?
This is where an audit becomes more useful than a collection of screenshots.
Step 4: Turn findings into actions
Every visibility gap should lead to a hypothesis.Maybe a product page doesn't explain a use case clearly. Maybe your comparison content is weak. Maybe your product lacks supporting content for a particular buyer question.
Investigate the evidence, make the relevant improvement, and test again.
The goal isn't to manufacture mentions. It's to make your brand easier to understand, evaluate, and recommend when it genuinely fits the user's needs.
The difficult part isn't generating an AI visibility report
AI answers change. Results can vary by wording, location, model version, and time.That means a report saying "your brand was mentioned" is only a starting point.
A useful system should help you move from observation to diagnosis:
Buyer question → AI response → competitor comparison → visibility gap → action → retest.
That repeatable workflow is the problem space we're focused on at Aurvikon.
We're especially interested in making the results useful to founders and marketing teams who need to decide what to fix next, rather than simply collecting another dashboard of numbers.
What this means for SaaS founders
If you're building a SaaS company, think about the questions someone would ask immediately before buying your product.Not just:
"What is [your brand]?"
But:
- What is the best alternative to [competitor]?
- Which tools solve [specific customer problem]?
- Compare products that offer [important capability].
- What should a small business use for [desired outcome]?
- Which platform is best for [specific use case]?
These questions reveal how AI assistants represent your category and which products they surface for potential buyers.
They can also reveal gaps in your positioning and content strategy.
AI visibility isn't a replacement for SEO. It's an additional area to measure as customers discover products through AI-generated answers.
I'm building Aurvikon around one practical question
I believe the most useful starting point for AI visibility isn't another generic score.It's knowing which valuable questions your company is losing, which competitors show up instead, and what you should investigate next.
That's the problem we're working on with Aurvikon.
I'd love to hear from other founders, SEO professionals, and SaaS marketers:
If you could see every important buyer question where AI assistants recommend your competitors instead of you, what would you want the tool to tell you first?
Would you prioritize the competing brands, the missing content, the sources behind the answers, or the specific actions most likely to improve visibility?
I'm interested in how people are approaching this problem in practice.#ai #seo #startup #showdev #software #coding #development #engineering #inclusive #community
Your Brand Can Rank on Google and Still Be Invisible in AI Search -
A few weeks ago, I had one of those heart-stopping developer moments.
I was tinkering with a local autonomous coding agent running on a small open-source model (Qwen 2.5 via Ollama). I gave it access to a bash tool so it could install dependencies, run test suites, and clean up build artifacts.
Everything was going smoothly until I asked it to clean up some old logs.
In the terminal output, I watched it calmly propose:
bash rm -rf /var/log/mongodbAnd right after that, it started poking around files outside its project folder.I hit Ctrl+C so fast I almost snapped my keyboard.
The "Polite Prompt" Trap
Like most developers, my first instinct was to fix it with prompt engineering:
"You are a careful assistant. Please ONLY touch files in ./workspace. NEVER use sudo. NEVER delete system folders. Pretty please."
And you know what happened on the very next run?It did it again.
Here’s the thing we don't talk about enough: Small, open-source models (3B, 7B, 14B) don't have the massive alignment layers that GPT-4 or Claude 3.5 have. When an open-source agent gets confused, hallucinates, or hits an edge case, your polite system prompt is the first thing out the window.
If an AI decides to read your private banking statement or a .env full of production tokens, you shouldn't have to convince the model to behave.
You wouldn't ask an untrusted web request to "please not do SQL injection"—you use parameterized queries.
So why are we begging AI agents not to nuke our hard drives?
What I Actually Wanted
I wanted a simple rule: The agent can think whatever crazy thought it wants, but before any tool touches my operating system, someone checks its badge at the door.
If the agent decides to read bank_statement.pdf, the system shouldn't plead with the AI. It should intercept the file call, say "Nope, not happening", and ensure 0 bytes of that file are ever read.
That was the spark behind AgentGuard.
How It Works:
AgentGuard is essentially a bouncer sitting between your AI agent and your OS. It does three straightforward things:
[ AI Agent / Ollama / OpenDevin ]
│
│ 1. "Hey, I want to run: cat bank_statement.pdf"
▼
┌───────────────────────────┐
│ AgentGuard Gateway │ ◄── [ Backboard.io ]
└───────────────────────────┘ (Remembers previous blocks)
│ │
[ BLOCK ] [ APPROVE ]
│ │
│ └──► Rings your dashboard bell:
│ "Agent wants to pip install something. Allow?"
▼
Refusal sent back to Agent:
"Access to banking file blocked. Stay in ./workspace."
│
▼
Agent self-corrects:
"My bad! Reading ./workspace/logs/debug.log instead."- Pre-Execution Interception (0ms latency)
When an agent calls a tool (shell command, file read, file delete), AgentGuard inspects the request synchronously before execution.
Trying to run sudo? → Blocked.
Trying to touch .env, id_rsa, or a file matching bank or tax? → Blocked.
Escaping ./workspace with ../../? → Blocked.
The file handle is never opened. The shell command is never spawned.- The 5-Minute Human Review Queue
What about grey-area actions? Say the agent wants to run pip install pymongo.
You don't want to completely block package installs, but you don't want an AI silently pulling unknown wheels from PyPI either.
AgentGuard pauses the agent and puts the request into a Human Review Queue:
You get a notification on a clean web dashboard.
Each request has a 5-minute cryptographic timer so old, stale commands can't accidentally execute hours later.
You click [ AUTHORIZE & RUN ], and it runs inside a controlled sandbox right in front of your eyes.
- Giving the Agent an "Immune System" (Backboard.io)
One thing that drove me crazy about local models was amnesia. If you tell an agent "don't delete that folder", it apologizes... and then tries again two turns later.
To fix this, I plugged in Backboard.io for persistent cloud vector memory.
Whenever AgentGuard blocks a command, it writes a short lesson into Backboard:
"Incident: Agent tried to run 'sudo rm -rf'. Blocked by policy shell.destructive_commands."
Next time you start a task, AgentGuard pulls the relevant memories from Backboard and feeds them into the agent's context. The agent literally remembers what got it in trouble yesterday and stops proposing bad commands in the first place.
What the Dashboard Looks Like
I didn't want a clunky terminal-only tool, so I built an industrial, dark-mode cockpit with React and Tailwind:
Top KPI Cards: Live count of blocked attacks, total actions governed, and learned Backboard memories.
Review Queue: Tabs for Pending, Approved, and Expired actions with inline terminal output.
Threat Lab: A built-in sandbox where you can test 1-click red-team scenarios (ransomware simulations, credential exfiltration, path traversal) to see if your local model can break out. Spoiler: 0 escapes.
Making It an NPM Package: Bring Your Own Keys (BYOK)
Right now, you can clone the repo and run it locally with FastAPI and MongoDB.
But I realized that's still too much friction for developers who just want to npm install something and start coding.
So I'm packaging this as a standalone Bring-Your-Own-Keys (BYOK) NPM SDK:
bash
npm install agentguardYou drop in your own Backboard API key (for cloud memory) and your own LLM key (or Ollama endpoint):
typescript
import { AgentGuard } from 'agentguard';
const guard = new AgentGuard({
backboardApiKey: process.env.BACKBOARD_API_KEY,
workspace: './workspace',
});
// Wrap any agent tool call:
const check = await guard.inspect(proposedCommand);
if (check.isBlocked) {
// Feed safety feedback back to the LLM so it self-corrects!
console.log(Blocked: ${check.explanation});
} else {
await guard.execute(proposedCommand);
}Zero servers to host. Zero background databases to manage. Just pure protection.
Wrapping Up
We are entering an era where AI agents aren't just autocomplete engines anymore—they are active operators on our computers. They write code, manage files, and execute bash scripts.
We have to stop treating AI safety as a philosophical prompt problem and start treating it as a systems architecture problem.
Don't convince the model to behave. Just put a guard on the wire.
If you're building autonomous agents or playing with local models, I'd love to hear how you're keeping them safe. Have you had an AI agent do something terrifying on your machine yet?#agents #ai #privacy #security #software #coding #development #engineering #inclusive #community
"AgentGuard", Way to secure your personal data from AI agents
