Transcript
00:00:00Claude Code, Codex and pretty much every other agent you use all have one serious problem and you run into it the moment you need one of them to actually make a decision.
00:00:08These models can think a problem through but they never come up with anything new.
00:00:12And it doesn't matter which one you're using because all of them fall short the moment you need something creative.
00:00:17So when you go to them for ideas, what comes back is the safe answer and you end up asking again and again for the angles they should have found on their own.
00:00:24Until you hand them those angles yourself, they won't think creatively at all.
00:00:28But the fix for this, weirdly enough, is actually ADHD which is when a person's attention jumps all over the place instead of staying on one thing.
00:00:35And it turns out ADHD becomes a superpower when you're an AI agent and that's basically what we're giving them now.
00:00:41There's a tool trending these days that does exactly that.
00:00:44And if this is your first time here, we're a software company and this is our channel AI Labs where we show you how to optimize your processes with AI just like we've optimized our own.
00:00:54So in this video, we're going to go over how giving your agent ADHD actually helps.
00:00:59Now before we get into the tool, you need to know why you need it in the first place.
00:01:03If you've worked with Claude Code, Codex or any other agent, you already know they're good at breaking the big tasks you give them into small chunks.
00:01:10They hand those chunks to sub-agents so each one works in its own context window without filling up the main session.
00:01:16You don't even have to ask for it.
00:01:18They create to-dos and delegate them wherever they can so the work runs in parallel.
00:01:22But all of that splitting up only happens for the work itself.
00:01:26None of it happens when you ask them to ideate and if you've tried brainstorming different angles of an idea with one of them, you'd already know this pain.
00:01:34You ask for variations and what comes back sounds different when you read it, but all of them are just the same idea reworded differently.
00:01:40So ideation is the trickiest thing to hand off because finding genuinely different directions is not something agents are good at.
00:01:47Basically, it comes down to how these models are trained.
00:01:50An agent reaches for the pattern that showed up most often in the data it learned from because seeing the same answer repeated is what taught it that the answer is a good one.
00:01:59That doesn't make its answers wrong and what you get is usually fine, but there are other angles to the problem it never even looks at, which kind of defeats the whole point of brainstorming with it.
00:02:08And the safe patterns are only half of it because it also never evaluates each idea on its own.
00:02:13Every possibility gets worked through in the same context window, so the ideas bleed into each other and the context fills up with noise.
00:02:20The thinking gets worse instead of better, that's why it can't evaluate anything clearly and it ends up handing you the same idea with different wording.
00:02:28The ADHD skill came out fairly recently and picked up a lot of stars within a few days.
00:02:33And the name fits because just like actual ADHD where the thoughts are scattered all over the place, it scatters the thinking instead of running it in one straight line.
00:02:42So it takes ideation or really any task with a lot of possible directions and splits it across separate agents as a tree of thought.
00:02:49A tree of thought is basically a structure where multiple branches of sub-agents work in isolation on different ideas, the same way a tree's branches split apart.
00:02:58Then at the end, once each branch has settled on one possibility, it brings all those ideas back together and merges them into the final answer.
00:03:06So it spins up multiple agents that think in their own context windows and each one gets a different framing of the same problem.
00:03:12They're isolated, so they share zero context and none of them know what the others are working on.
00:03:17But the isolation here isn't about splitting up the work, it's about keeping each idea separate so they can't influence each other.
00:03:24The way it keeps them apart is with frames which are basically different lenses to look at the problem through.
00:03:29It has a whole frame library inside it which holds all the different directions the agents can go in and each agent picks the frame it's going to work through.
00:03:37So it gets that frames prompt along with the system prompt and the problem itself.
00:03:41Then a critic agent scores everything that comes back because something has to decide which of these ideas are worth keeping.
00:03:48It grades every idea on three things.
00:03:50Novelty, which is how new and creative the idea actually is.
00:03:54Viability, which is whether you could realistically build it.
00:03:57And fit, which is how well it matches the problem you're trying to solve.
00:04:01This grading happens in an agent of its own, running a prompt that tells it to act as a skeptical senior engineer.
00:04:07So its whole job is to be hard on everything it reads.
00:04:10And based on the score, it decides whether an idea survives or gets thrown out.
00:04:15At the end, the skill shortlists the strongest ideas and goes through the trap list which is basically the problems each idea could cause if you actually went with it.
00:04:23Then it prioritizes those ideas that got flagged as non-obvious.
00:04:26Before we actually show you interesting ways this skill actually helps, it would be great if you subscribe to the channel and hit the hype button.
00:04:33This small gesture of support goes a long way for us.
00:04:36So that's what it's doing under the hood.
00:04:38Now let's get it installed.
00:04:40You'll find the command for that on the project's GitHub repo.
00:04:43So you copy it, open the terminal in whichever project you're working on and run it.
00:04:47It then asks which AI coding agent you want it installed for, and it supports more than 45 of them.
00:04:53So you can just pick whichever one you use.
00:04:55After that, it asks whether it should be available only inside the current project, which is Project Scope, or from anywhere, no matter which project you're in.
00:05:03If you only need it in one place, Project Scope is the one to pick.
00:05:06But once it's done, the skill lands in a folder called .agents.
00:05:10That's what a lot of other agents like Codex use for keeping their configuration, but Claude code only recognizes the .clawed folder by default.
00:05:18So if that's what you're using, you rename that folder to .clawed and it gets picked up.
00:05:22When you open it, you'll see the skill sitting in there.
00:05:25And it's just a single skill.md file that handles everything by itself, with no reference files or dependencies next to it.
00:05:31The instructions also push the agent hard past its first three answers.
00:05:36The file straight up says those first three are the most common responses in the data these models learned from,
00:05:41and they're also the ones any senior agent could come up with immediately.
00:05:44The more interesting stuff a senior agent is actually capable of only starts after those.
00:05:50But spinning up this many agents burns a lot of tokens,
00:05:52so there's a pre-check step that decides whether this skill should run at all.
00:05:56If you call it with the slash command, or you just ask for it directly, it fires straight away.
00:06:01If you didn't mention it in your prompt to call the skill directly,
00:06:04and your agent decides to auto-invoke the skill,
00:06:07then it runs the problem through three questions that makes up the pre-check step.
00:06:11The first one is whether the problem is open-ended,
00:06:14which basically means, would someone experienced have a few different answers that all work here,
00:06:19or just one right answer.
00:06:20If there's only one right answer,
00:06:22then thinking from multiple angles is pointless and just wastes tokens,
00:06:26so it stops there.
00:06:27The second one is whether the stakes are actually high,
00:06:29which means, would it genuinely cost you something if the obvious answer turned out to be the wrong one.
00:06:34And the third one is how you asked.
00:06:36If you used words like quick or standard,
00:06:38you're clearly after the straightforward answer,
00:06:40so it stops there instead of using the skill.
00:06:43Asides from the pre-step,
00:06:44you've got all the phases of the loop along with the table of frames which become the different directions that get handed to each agent.
00:06:50It also lists the patterns the agents are explicitly told to avoid.
00:06:54But before we dive into interesting ways you can use it, let's have a word by our sponsor.
00:06:59Top View.
00:06:59If you make AI videos, you already know the pain.
00:07:02Every model lives on a different platform and you generate one clip at a time.
00:07:05Top View fixes that.
00:07:06It's the world's first all-in-one AI video skill and it lives right inside your coding agent.
00:07:11Claude code, cursor, codex.
00:07:13It aggregates every top model in one place.
00:07:16VO, Kling, Sea Dance, Nano Banana and more.
00:07:19No switching platforms, no juggling subscriptions.
00:07:22So we tried it.
00:07:23We gave our agent one command.
00:07:25Generate 10 variations of a 15-second TikTok ad from this product image.
00:07:29Seconds later, we had 10 finished ads ready to post.
00:07:32This is the real shift.
00:07:34Top View turns your agent into a video production line.
00:07:36You describe what you want once and it batch generates dozens of videos across different models, styles and aspect ratios in a single session.
00:07:44It even auto-selects the right model for the job.
00:07:47You go from one line of text to finished videos without ever leaving your agent.
00:07:51Try Top View's skill at the link in the description.
00:07:54So that's its setup.
00:07:54Now let's get into where this actually pays off.
00:07:57One place it really pays off is test-driven development, which is also called TDD, where you have the agent write the tests first.
00:08:03Then it builds the app piece by piece until all of those tests pass.
00:08:07And writing the tests before you write the code matters, like we've talked about in our previous videos,
00:08:12because when all of your requirements are strictly written out as code, any change that breaks the app gets caught by the tests.
00:08:18The agent is forced to comply with them.
00:08:20And writing tests is a really good problem to hand to the skill, because this is exactly where agents slack off.
00:08:26They don't cover all the cases they should, because like we just said, they fall back on the same common answers and only write tests around those.
00:08:33They never look at the other paths someone could take through the app, which need to be covered too.
00:08:37You can hand one a detailed prompt on how to write tests, and you can build a specialized test author agent whose only job is writing them.
00:08:45But they still fall back to the same patterns.
00:08:47Before you run it though, the agent needs to know what you're building before it writes anything.
00:08:51For that, you need to write down what needs to be built, like a PRD, which is basically the document that lays out what the app is supposed to do and the problem it solves, along with the goals you want to hit and who it's for.
00:09:03And alongside that, you should also give it a technical specification document, which locks in the technical details so you're not repeatedly telling it which tools to use.
00:09:11You link both of those inside your Claude.md file, so it picks up that context from the very start.
00:09:16Then you invoke the skill with its slash command, give it a prompt describing the app you're building, and ask it to write test cases using a TDD approach.
00:09:25Because you called it explicitly, it skips the pre-check and spins up five agents straight away.
00:09:30Each one goes off in its own direction of thinking, using the frame that best matches the problem, and they come back with different approaches for writing the tests.
00:09:38Then it scores each one against the criteria we went through earlier, picks the top three, and explores those in more depth.
00:09:44And once all of those agents have finished, you get a detailed report of the testing directions along with their scores.
00:09:50The scores are written in shorthand, so N9 means the idea scored a 9 on novelty, V8 means it scored an 8 on viability, and F10 means it got a perfect score on fit.
00:10:00Each idea also comes with a sketch of how it would be built, the risks involved, and the first steps for getting started on it.
00:10:07But the ideas that come back look nothing like a normal set of tests.
00:10:11Usually the tests an agent writes only check that the app works correctly.
00:10:15But the ideas the skill gave went after way more of the edge cases, and they catch performance issues too.
00:10:21So you end up with a much stronger test suite, split across three deeply explored branches, and each one is testing a different path through the application.
00:10:30One thing to be clear about here though, the skill plans the tests, it doesn't actually write them.
00:10:34What you get back is the strategy, so from there you just tell the agent which direction you want, and it goes and implements that one.
00:10:40You can take a single direction if that's all you need, or have it implement all three.
00:10:44If performance is critical for what you're building, you'd want all three, but that takes a while, because the agent has to work through the paths one at a time.
00:10:52Once it's done though, you can clearly see the tests are way more detailed than they would have been otherwise,
00:10:57because the whole testing strategy got planned out in depth before a single test was written.
00:11:02So this works really well before you start building, since it covers most of the ground up front, and cuts down the chances of breaking the app later on.
00:11:09But that's all for before you build.
00:11:11The other way to use this is as a step right before you ship.
00:11:14You run the skill on the app you're about to launch, and ask it to evaluate the user experience.
00:11:18Then it flags anything that could trip people up while they're navigating the site or using the product,
00:11:23and anything that could push them to churn, which is basically when people stop using your product after they've already started.
00:11:29And churn happens a lot once your site is out there with real people using it, especially if it's a paid one.
00:11:34They like it at first and then leave, because one feature didn't work the way they expected,
00:11:38so they want their money back.
00:11:40And a lot of the time, it's something small that got overlooked while the app was being built,
00:11:44made its way into the live version, and caused problems down the line.
00:11:48We ran this on our own community website, and we were actually in the middle of launching a new feature.
00:11:53We already have a lot of members in there, so anything new has to be checked really carefully,
00:11:57because we don't want to ship something that breaks the experience for people who are already there.
00:12:01So we invoked it with the slash command, gave it the feature, and asked it to work out
00:12:05where people might churn and what could give them a bad experience.
00:12:09It started by going through the application in depth to gather context,
00:12:12then spun up its agents the same way and surfaced around 30 different ideas.
00:12:17From those, it picked the top three and explored them further.
00:12:20And once that was done, it gave us every finding graded on novelty, viability, and fit.
00:12:25And it caught gaps that had gone completely undetected,
00:12:28like features that were promised in the PRD but never actually got built,
00:12:32so we'd have shipped something that didn't match what we said it would do,
00:12:35along with a bunch of other findings.
00:12:37And for every one of them, it suggested a fix and listed out the traps and risks that came with it.
00:12:42And it works the same way as the tests here.
00:12:44It's not fixing anything itself, so you just hand the findings you want back to the agent,
00:12:48and it goes and implements them.
00:12:50So you get to deal with all of that before the launch instead of after it,
00:12:53which puts your app in a way better state by the time it actually goes public.
00:12:57Now all the skills, workflows, and resources we show you in our videos
00:13:01are available in AI Labs Pro, which is our community.
00:13:04So if you've found value in what we do and want to support the channel,
00:13:07this is the best way to do it.
00:13:09The link's in the description.
00:13:10That brings us to the end of this video.
00:13:12If you'd like to support the channel and help us keep making videos like this,
00:13:16you can do so by using the super thanks button below.
00:13:18As always, thank you for watching, and I'll see you in the next one.