1 00:00:02,619 --> 00:00:08,003 [music] Hi, welcome to another video. 2 00:00:08,003 --> 00:00:12,709 So, there's this new Claude Code plugin that has been blowing up on GitHub recently. 3 00:00:12,719 --> 00:00:21,348 It's called I have ADHD, and it has already crossed about 16,000 stars, which is pretty insane for something that's essentially just a single skill file. 4 00:00:21,348 --> 00:00:26,432 Now, before you ask, no, you don't need an actual ADHD diagnosis to use this. 5 00:00:26,432 --> 00:00:35,898 The whole point of this plugin is that it fixes something that annoys literally everyone who uses AI coding assistants, and that's the fact that these models just talk way too much. 6 00:00:35,898 --> 00:00:46,115 You ask a simple question, and the AI gives you three paragraphs of preamble, a lecture on best practices, five tangential suggestions you didn't ask for, and then ends with, "Hope this helps. 7 00:00:46,115 --> 00:00:52,986 Let me know if you have any questions." And somewhere in the middle of all that buried like treasure, is the actual answer you wanted. 8 00:00:52,986 --> 00:00:59,110 So, this plugin basically reformats all of Claude's responses to be action first, numbered, and completely free of filler. 9 00:00:59,120 --> 00:01:02,254 And to be honest, after using it for a bit, it's kind of hard to go back. 10 00:01:02,254 --> 00:01:07,510 So, let's get right into it and see what it actually does, how to install it, and whether it's worth adding to your setup. 11 00:01:07,520 --> 00:01:10,085 Now, let's talk about what this thing actually is. 12 00:01:10,085 --> 00:01:14,101 I have ADHD is a skill for coding agents made by a developer called Igri. 13 00:01:14,101 --> 00:01:19,846 It works with Claude code as a plugin, and there are also instructions for using it with other agents like Codeex. 14 00:01:19,846 --> 00:01:25,475 The whole thing is open source under the MIT license and it's completely free which is always great to see. 15 00:01:25,475 --> 00:01:28,003 The interesting part is the thinking behind it. 16 00:01:28,003 --> 00:01:35,271 The skill is loosely inspired by an actual book called the ADHD toolkit but adapted for how LLM communicate. 17 00:01:35,271 --> 00:01:39,190 The idea is that people with ADHD have limited working memory. 18 00:01:39,200 --> 00:01:45,698 They struggle with the gap between understanding something and actually doing it and they need visible progress to stay motivated. 19 00:01:45,698 --> 00:01:52,389 And if you think about it, that's basically everyone when they're tired, distracted, or just trying to ship something at 2:00 a.m. 20 00:01:52,399 --> 00:01:59,340 So, the skill takes those principles and turns them into 10 rules that the AI has to follow in every response. 21 00:01:59,340 --> 00:02:04,069 Now, let me quickly run you through the main rules because this is where it gets good. 22 00:02:04,079 --> 00:02:06,798 First, the AI has to lead with the action. 23 00:02:06,798 --> 00:02:15,994 So instead of explaining why authentication is complex, it just tells you to run the npm install command and edit the specific file at the specific line. 24 00:02:15,994 --> 00:02:17,797 No preamble at all. 25 00:02:17,797 --> 00:02:23,915 Second, every multi-step task gets numbered and each step is one single bounded action. 26 00:02:23,915 --> 00:02:26,070 No and then chains hidden inside a step. 27 00:02:26,080 --> 00:02:32,134 Third, every response has to end with one concrete next action that you can do in under two minutes. 28 00:02:32,134 --> 00:02:37,102 This is actually really clever because the hardest part of any task is just starting it. 29 00:02:37,102 --> 00:02:44,549 Then there's tangent suppression, which means the AI finishes the main issue first and only offers secondary stuff as separate follow-ups. 30 00:02:44,549 --> 00:02:48,183 There's also a rule where it restates the state on each turn. 31 00:02:48,183 --> 00:02:52,856 So it'll say something like step three of five done schema updated. 32 00:02:52,856 --> 00:02:59,424 Next is backfilling the column so you never lose track of where you are in a long session, which is honestly useful for everyone. 33 00:02:59,424 --> 00:03:08,442 It also has to give specific time estimates like about 15 minutes if tests cover this and afternoon if not instead of vague stuff. 34 00:03:08,442 --> 00:03:12,120 Errors get reported matterof factly with the cause and the fix. 35 00:03:12,120 --> 00:03:14,343 No uh-oh or nervous hedging. 36 00:03:14,343 --> 00:03:16,895 Lists are capped at five items maximum. 37 00:03:16,895 --> 00:03:21,463 And finally, no preamble, no recaps, and no closing pleasantries. 38 00:03:21,463 --> 00:03:24,762 So no more let me know if this helps at the end of every single message. 39 00:03:24,762 --> 00:03:27,910 There are also some sensible overrides built in. 40 00:03:27,920 --> 00:03:32,149 If you ask for an explanation, it'll still explain properly with structure. 41 00:03:32,159 --> 00:03:34,823 If something is destructive, it'll confirm first. 42 00:03:34,823 --> 00:03:40,710 And if your request is genuinely ambiguous, it'll ask one clarifying question instead of guessing. 43 00:03:40,720 --> 00:03:42,198 clarifying question instead of guessing. 44 00:03:42,198 --> 00:03:45,204 So, it's not just blindly tur, which I appreciate. 45 00:03:45,204 --> 00:03:47,904 Now, let's install it and see it in action. 46 00:03:47,904 --> 00:03:49,715 The installation is pretty simple. 47 00:03:49,715 --> 00:03:54,630 Run claude plug-in install I have a HD at I have a HD. 48 00:03:54,640 --> 00:03:55,135 That's it. 49 00:03:55,135 --> 00:03:56,025 You're good to go. 50 00:03:56,025 --> 00:04:01,249 Once it's installed, you just open Claude Code and type /avad HD to activate it. 51 00:04:01,249 --> 00:04:06,710 And if you ever want to turn it off, you can just run claude plugin disable I have a HD. 52 00:04:06,720 --> 00:04:09,440 So I have Claude code open here with the skill activated. 53 00:04:09,440 --> 00:04:11,077 Let me show you the difference. 54 00:04:11,077 --> 00:04:13,341 First, let me ask it something without the skill. 55 00:04:13,341 --> 00:04:17,518 Just as a baseline, I'll ask how to add JW2 authentication to my Express app. 56 00:04:17,518 --> 00:04:24,061 And you can see it starts with a whole explanation of what JWT is, why authentication matters, the different approaches you could take. 57 00:04:24,061 --> 00:04:25,070 It's a wall of text. 58 00:04:25,070 --> 00:04:28,204 The actual commands are in there somewhere, but you have to dig for them. 59 00:04:28,204 --> 00:04:30,677 Now, let's try the same thing with the skill active. 60 00:04:30,677 --> 00:04:32,070 Let's send it and see. 61 00:04:32,080 --> 00:04:32,456 Okay. 62 00:04:32,456 --> 00:04:33,736 And look at this. 63 00:04:33,736 --> 00:04:36,840 The very first line is the install command. 64 00:04:36,840 --> 00:04:39,316 Then it gives me four numbered steps. 65 00:04:39,316 --> 00:04:43,185 Each one a single action with the exact file and line number to edit. 66 00:04:43,185 --> 00:04:48,340 At the end, it says the next action is to run the dev server and test the login route. 67 00:04:48,340 --> 00:04:51,003 And it estimates the whole thing at about 20 minutes. 68 00:04:51,003 --> 00:04:53,038 That's the entire response. 69 00:04:53,038 --> 00:04:56,390 No fluff, no lecture, no great question. 70 00:04:56,400 --> 00:04:58,390 This is pretty amazing to be honest. 71 00:04:58,400 --> 00:05:01,045 Now, one more thing that's worth mentioning is customization. 72 00:05:01,045 --> 00:05:11,538 Since the whole skill is just a single skill.md file, you can fork the repo, edit the file at skills/avadh/skill.md, and change whatever you want. 73 00:05:11,538 --> 00:05:16,640 Maybe you want lists capped at three items instead of five, or you want it even more aggressive with the brevity. 74 00:05:16,640 --> 00:05:22,870 You just edit the file, rerun the slash command, and your personalized version is active. 75 00:05:22,880 --> 00:05:28,884 The repo has also been translated into a bunch of languages like Chinese, Japanese, Korean, and Portuguese. 76 00:05:28,884 --> 00:05:29,831 So, that's cool, too. 77 00:05:29,831 --> 00:05:31,162 So, what's my verdict on this? 78 00:05:31,162 --> 00:05:34,769 I think this is one of those ideas that's obvious in hindsight. 79 00:05:34,769 --> 00:05:40,841 The models have gotten really good at coding, but the way they communicate is still tuned for politeness rather than usefulness. 80 00:05:40,841 --> 00:05:44,050 And this skill fixes that with basically zero cost. 81 00:05:44,050 --> 00:05:44,710 It's free. 82 00:05:44,720 --> 00:05:45,555 It's open source. 83 00:05:45,555 --> 00:05:48,226 It's a two-minute install, and you can turn it off anytime. 84 00:05:48,226 --> 00:05:54,751 If you ask me, the progress restating and the end with one small next action rules are the real killers here. 85 00:05:54,751 --> 00:05:58,638 Those alone make long agentic sessions way easier to follow. 86 00:05:58,638 --> 00:06:03,339 The token savings from cutting all the filler are a nice bonus as well, especially if you're on a limited plan. 87 00:06:03,339 --> 00:06:05,173 There are some downsides, of course. 88 00:06:05,173 --> 00:06:12,911 If you're a beginner who actually wants the explanations, this might feel too tur and you'd need to explicitly ask it to explain things. 89 00:06:12,911 --> 00:06:19,175 But for anyone who already knows what they're doing and just wants answers, this is a really good option for sure. 90 00:06:19,175 --> 00:06:20,876 Overall, it's pretty cool. 91 00:06:20,876 --> 00:06:23,364 Anyway, let me know your thoughts in the comments. 92 00:06:23,364 --> 00:06:29,137 If you like [music] this video, consider donating through the super thanks option or becoming a member by clicking the join button. 93 00:06:29,137 --> 00:06:32,309 Also, [music] give this video a thumbs up and subscribe to my channel. 94 00:06:32,319 --> 00:06:33,608 I'll see you in the next one. 95 00:06:33,608 --> 00:06:47,404 Until then, bye. 96 00:06:47,414 --> 00:06:49,434 >> [music]