Developer Builds Claude Code Skills to Compensate for Executive Function Challenges

How I Use AI as an Executive Function Prosthetic

For years I thought I had a discipline problem. I had shipped code, finished a degree, built things, and still the dominant private feeling was that I was getting away with something, that the gap between what I could do on a good day and what I could do on a normal one was a character flaw I was hiding. The reframe that changed everything was clinical, not motivational: I do not have a discipline problem. I have an executive function problem. And executive function, unlike character, can be supported from the outside. This is the most personal of the three posts in this cluster. The other two are practical: the CLAUDE.md guide and the five skills. This one is the why underneath both. What Is Executive Function, Actually? Executive function is the brain's management layer. It is not intelligence, and it is not knowledge. It is the set of processes that turn knowing-what-to-do into actually-doing-it. The National Institute of Mental Health describes ADHD as fundamentally a disorder of these self-management processes rather than of attention alone. It is not one function. For the purposes of getting work done, it is at least four distinct ones, and ADHD disrupts each of them in a different way: Working memory, the mental scratchpad holding what you are doing right now. Task initiation, the ability to start, to cross the gap from intention to action. Context switching, the ability to drop one task, pick up another, and come back without losing the first. Time perception, the internal sense of duration that lets you pace and estimate. Calling them out separately matters, because "I struggle with executive function" is too vague to act on. Each of the four breaks differently and each one needs a different prosthetic. Lumping them together is how you end up trying to fix a time-perception problem with a task-initiation strategy and concluding you are just broken. What Is an Executive Function Prosthetic? A prosthetic does not heal. It compensates. Glasses do not repair your eyes, they do the focusing your eyes cannot, and the result is that you see. Nobody calls glasses a moral failure. Nobody tells a nearsighted person to try harder at focusing. An executive function prosthetic is the same idea pointed at the management layer. It is an external system that performs the executive work your brain struggles to perform, so the outcome is the same as if the function worked. A calendar is a prosthetic for time perception. A checklist is a prosthetic for working memory. The framing I want you to take from this post is that you are allowed to build these, deliberately, for the specific parts of you that do not work, and that doing so is engineering, not cheating. The crutch objection answers itself. Yes, it is a crutch. Crutches let people walk who otherwise could not. The point of an assistive tool was never to prove you can do it the hard way. It was the walking. I use Claude Code as a prosthetic for all four functions. Here is the map. How ADHD Disrupts Each Function, and How AI Compensates Working Memory How ADHD breaks it: the scratchpad is smaller and leakier. You hold fewer items and they fall off faster, especially across interruptions. Close a project, come back, and the model of what you were doing is partly or wholly gone. The prosthetic: I move the scratchpad out of my head into a file Claude reads automatically. My conventions, my stack, my current task, all of it lives in CLAUDE.md, so the cost of dropping it from memory goes to zero. The file remembers, so I do not have to. The first time I reopened a project and Claude told me what I was doing before I said a word, the relief was physical. Task Initiation How ADHD breaks it: there is a real barrier between planning and starting. It is not laziness, it is a failure of the initiation circuit to fire, and a blank editor is the worst possible trigger because it offers nothing to grab. You can know exactly what to do and still be unable to begin. The prosthetic: I never start from a blank slate. I describe the goal and Claude decomposes it into three to five concrete tasks, and I pick one. Picking one item off a list is a completely different cognitive act than generating the first move from nothing, and the initiation circuit, which could not fire on "build the feature," fires easily on "do this one named thing." The decomposition is the prosthetic. The list is the ramp. Context Switching How ADHD breaks it: every switch tears down the mental model, and rebuilding it costs time. The American Psychological Association measures the typical recovery at about 23 minutes per switch. ADHD switches more often, involuntarily, and rebuilds slower. Ten interruptions is not ten minutes lost, it is most of a working day bled out in reconstruction. The prosthetic: the reconstruction gets done by the tool, not by me. At session start, Claude reads the checkpoint and summarizes the open state, what was in flight, what was blocked, before I write a line. The cold start becomes a warm one. The 23 minutes I would have spent opening files semi-randomly trying to jog the memory get spent on the actual work instead. Time Perception How ADHD breaks it: you cannot feel time passing. "One quick fix" becomes four hours. A two-hour task gets estimated at twenty minutes. You ship on deadline panic because there is no internal clock to pace against, and the absence is invisible from the inside, which is what makes it so hard to fix by awareness. The prosthetic: external time markers, generated by the tool, surfaced at the moment of drift. Checkpoints at real intervals that interrupt and ask "you set 45 minutes, you are at 60, still the right task?" The brain that cannot generate the time signal gets handed one, while the drift is happening rather than after. The clock lives outside me because inside me it does not run. The One That Is Not on the Standard List: Discounting the Positive There is a fifth gap that is not technically executive function, but it sabotages all four, so I treat it as part of the same system. "Discounting the positive" is a cognitive distortion from CBT: you mentally erase your own wins, so only the failures register. You shipped three features and your brain renders the one bug. ADHD ships with this so reliably that it stops feeling like a distortion and starts feeling like accuracy. It is not accuracy. It is a measurement error, and it is expensive, because the ADHD reward system is already under-fueled, and a brain that cannot see its own progress gets none of the dopamine that progress is supposed to pay. This is where my mirror skill came from. It is the most personal thing I have built. At the end of a session, it reflects back what I actually did, grounded in evidence: the commits, the closed tasks, the things that shipped. Not encouragement, receipts. When I tell myself "good job," my brain discounts it in real time. When a tool shows me the commit log of what I closed today, the distortion has nothing to argue with. The win gets to count. That last clause is the whole reason this matters. For an under-fueled reward system, letting the win count is not a nice-to-have. It is the fuel that makes tomorrow's initiation possible. Externalizing the scorekeeping is how I keep the distortion from quietly stealing the dopamine I need to keep going. What This Looks Like in Practice A single feature, start to finish, mapped to the four functions plus the fifth: Intention: "Add rate limiting to the API." [Task initiation] I do not stare at a blank editor. I hand Claude the goal. It returns: 1) define the limiter, 2) wire the middleware, 3) add the headers, 4) write the tests. I pick #1. [Working memory] Everything I decide (Redis-backed, 100 req/min, per-key) goes straight into CLAUDE.md. I do not hold it. It holds it. [Context switch] Slack pulls me out for 40 minutes. I come back cold. Claude reads the checkpoint: "limiter defined, middleware next." No 23-minute rebuild. Warm start. [Time perception] A 45-minute checkpoint pings at 60. "Still on the limiter?" I had drifted into refactoring an unrelated helper. Caught it now instead of two hours from now. [Discount-positive] End of session my brain says "you barely did anything." /mirror: "Shipped rate limiting, 4 tasks closed, tests green. Commits attached." The win counts. Fuel for tomorrow. No single piece of that is heroic. The system is the point. Each prosthetic is dull on its own. Together they let a brain that cannot reliably do any of the four functions produce the same output as one that can. The Honest Limits This is environmental support, not treatment. It does not replace medication, therapy, or a clinician, and I would not want anyone to read a blog post and substitute it for care. The prosthetic frame is a way to design your environment, not a way to fix your neurology. It also has failure modes. A prosthetic that lies is worse than none: stale context in CLAUDE.md will make Claude reason confidently from a model that no longer matches reality, and you have to maintain the aid with the same discipline you lack, which is its own quiet joke. And there is a real risk of outsourcing understanding, of letting the tool carry so much that you stop being able to hold any of it. The guard there is that I still have to understand what ships. The prosthetic carries the management layer. It does not get to carry the thinking. But the reframe stands, and it is the thing I most want you to take. I am not undisciplined. My executive function works differently, and the parts that do not work can be supported from the outside, deliberately, without shame. I stopped trying to white-knuckle my way into a neurotypical brain. I started building prosthetics for the specific functions mine does not have. The work got done. So did I. Frequently Asked Questions What is an executive function prosthetic? An external system that performs the executive-function work your brain struggles to do, the way glasses do the focusing your eyes cannot. It does not heal the underlying difference. It compensates so the outcome is the same as if the function worked. Is using AI for executive function just a crutch? Yes, and that is not an insult. Crutches let people walk who otherwise could not. The goal of an assistive tool is the outcome, not proving you can do it the hard way. A calendar is a crutch. So are glasses. What is "discounting the positive"? A cognitive distortion from CBT where you mentally erase your own wins so only the failures register. It is common with ADHD and it starves an already under-fueled reward system, which is why externalizing your own scorekeeping helps. Does this replace ADHD medication or therapy? No. It is environmental support, not treatment. It works alongside medication, therapy, and other supports, never instead of them. Talk to a clinician about treatment and use tools like this to design your environment. If this mapped onto your experience, the two practical companions are the ADHD developer's guide to CLAUDE.md for the working-memory layer, the five Claude Code skills for the rest of the executive-function gaps, and the original ADHD coding workflow that started this whole cluster. Originally published on chudi.dev

6 hours ago

5 Claude Code Skills Every ADHD Developer Needs

I have built 114 Claude Code skills. Most of them are engineering plumbing. But five of them exist for one reason only: my executive function has specific, repeatable holes, and I got tired of falling into the same ones. These five are not productivity hacks. Each one maps to a named ADHD deficit, and each one fills it the same way every time so I do not have to re-improvise around my own brain at 2pm. If you want the broader system this sits inside, start with my Claude Code ADHD workflow and the CLAUDE.md guide. This post is the skills layer specifically. What Is a Claude Code Skill? A skill is a named, repeatable workflow you invoke with a slash command. Instead of re-prompting Claude Code from a blank slate every time ("okay, help me figure out what to work on, here is my situation again..."), you type /adhd-task-triage and it runs the same defined steps it ran yesterday. For an ADHD brain, that determinism is the feature. The skill does not depend on me remembering how to drive it. It just runs. Custom skills live in a .claude/skills/<name>/SKILL.md file that describes what the skill does and when it should fire. You can build one for any gap you fall into more than twice. 1. adhd-task-triage: Energy-Based Prioritization The gap it fills: task initiation paralysis. Standard task managers sort by priority or deadline. That assumes you can act on the top item by willpower. ADHD does not work that way. The top-priority task and the task you can actually start right now are often different tasks, and trying to force the high-priority one when your initiation circuit is offline produces zero output and a guilt spiral. adhd-task-triage sorts by available energy, not importance. You tell it where you are (wired, foggy, depleted), it looks at the work in front of you, and it hands back the task that matches the state you are actually in, not the one you wish you were in. /adhd-task-triage Why it helps specifically: it removes the moral framing. The question stops being "why can't you just do the important one" and becomes "what can this brain, in this state, actually initiate." Matching the task to the energy is how you get momentum, and momentum is the thing ADHD brains can ride once it exists. 2. mirror: The Counter to Discounting the Positive The gap it fills: cognitive distortion, specifically discounting the positive. "Discounting the positive" is a CBT term for a thinking pattern where you mentally delete your own wins. You shipped three features, but your brain only renders the one bug. ADHD comes bundled with this so often that it feels like just being realistic. It is not realistic. It is a measurement error, and it is corrosive, because a brain that cannot see its own progress loses the dopamine that progress is supposed to pay out. mirror reflects back what I actually did, grounded in evidence, not vibes. It reads the session, the commits, the closed tasks, and tells me plainly: here is what shipped. Not cheerleading. Receipts. /mirror Why it helps specifically: ADHD reward systems are under-fueled, and discounting the positive starves them further. When the skill says "you closed four tasks and shipped the auth flow," and points at the commits, the distortion has nothing to push against. I cannot argue with the log. The win gets to count, which is the entire point. The reason this works is that it is external and evidence-based. If I tell myself "good job," my brain discounts it instantly. If a tool shows me the commit history of what I closed today, the distortion has no grip. Externalize the scorekeeping and the distortion loses. 3. pickup: Session Resume for Context-Switching The gap it fills: context-switch recovery cost. Every interruption tears down the mental model of what I was doing. Rebuilding it is the 23-minute reconstruction tax that, for an ADHD brain, runs longer and hits more often. The worst version is the overnight gap: I close the laptop mid-thought and the next morning the entire working model is just gone. pickup reconstructs the session for me. It reads where I left off, what was in flight, what was blocked, and summarizes the open state before I write a single line. /pickup Why it helps specifically: it converts a cold start into a warm one. The reconstruction that used to eat the first half hour of every session, the part where I open files semi-randomly trying to jog the memory, gets done by the skill in seconds. I am not rebuilding the model. I am reading it back. 4. schedule: A Patch for Time Blindness The gap it fills: time perception distortion. Time blindness is an ADHD hallmark: you cannot feel time passing, so "I'll just fix one thing" becomes a four-hour rabbit hole, and a two-hour task gets estimated at twenty minutes. You ship on deadline panic because your brain has no internal clock to pace against. schedule puts external markers on the work: it sets up recurring checkpoints and time-boxed runs so the passage of time becomes visible instead of imagined. It is the external clock my brain does not have. /schedule Why it helps specifically: ADHD does not respond well to "be more aware of time," because the awareness machinery is the thing that does not work. It responds to external structure. A skill that interrupts at a real interval and asks "you set 45 minutes for this, you are at 60, still the right task?" gives me the time signal my brain cannot generate, at the moment the drift is happening rather than after. 5. librarian: Reducing the Cognitive Load of Navigation The gap it fills: working-memory overload from holding a whole codebase in your head. Navigating a large codebase means holding a map of it in working memory: where things are, what calls what, which file owns which concern. ADHD working memory cannot hold that map, so every navigation becomes a fresh, expensive search, and the expense itself becomes a reason to avoid touching unfamiliar parts of the code. librarian walks the knowledge layer for me. Instead of me grepping around trying to reconstruct where a behavior lives, I describe what I am after and it returns the relevant pieces and how they connect. It holds the map so I do not have to. /librarian Why it helps specifically: it collapses the navigation tax. The mental energy I would spend reconstructing "where does this live and what touches it" is exactly the executive-function budget I do not have to spare. Offloading the map means I can spend that budget on the actual change instead of on finding the place to make it. What This Looks Like in Practice A real morning, lightly compressed, showing the five working as a chain rather than in isolation: 9:10 Sit down. No memory of yesterday's state. /pickup -> "You were mid-refactor on the signal loop, JWT middleware blocked on a cookie setting. 2 tasks open." Cold start avoided. ~20 min reconstruction skipped. 9:14 Brain is foggy, not wired. /adhd-task-triage -> hands me the low-initiation-cost task (write the middleware test), not the hard one. Momentum started instead of staring at the blank editor. 9:20 Need to change the auth schema but forget where it lives. /librarian -> returns the schema file + everything that touches it. Navigation tax collapsed. No grep spiral. 9:25 /schedule already running a 45-min checkpoint in the background. At 10:10 it pings: "60 min on a 45 task, still the right one?" Caught a hyperfocus drift before it ate the morning. 11:30 Brain says "you got nothing done." /mirror -> "3 tasks closed, middleware shipped, schema migrated. Here are the commits." Distortion loses. The win counts. None of these is doing anything a disciplined neurotypical developer could not do by hand. That is the point. They do it for me, the same way every time, so the doing does not depend on executive function I cannot summon on command. How Do You Start Building Your Own? Do not install five skills today. Pick the one deficit that costs you the most, mine was the cold-start reconstruction, and build or invoke the one skill that targets it. A skill is a SKILL.md file in .claude/skills/ that names what it does and when it triggers. Start there, use it for a week, and let the next-most-expensive gap tell you what to build next. The pattern underneath all five is the same: name the executive-function gap precisely, then build a repeatable thing that fills it so you stop re-improvising around your own brain. Frequently Asked Questions What is a Claude Code skill? A packaged capability you invoke by name (like /pickup) that runs a defined, repeatable workflow. Instead of re-prompting Claude from scratch each time, you call the skill and it runs the same proven steps every time. Are these built into Claude Code? Claude Code ships with a skill system and some bundled skills. Several here (like adhd-task-triage and mirror) are custom skills I built for my own setup. The transferable thing is the pattern: build a skill for any recurring executive-function gap. Do I need ADHD for these to help? No. They help anyone who context-switches a lot, undersells their progress, or loses the thread between sessions. ADHD just makes the gaps sharper, so the payoff comes faster. How do I invoke a skill? Type its slash command (for example /pickup) in Claude Code. Custom skills live in .claude/skills/ as a SKILL.md file that defines what the skill does and when it fires. These skills are the executive-function layer on top of the working-memory layer from the CLAUDE.md guide. For the clinical picture of why each gap exists and how AI compensates, read how I use AI as an executive function prosthetic. Originally published on chudi.dev

6 hours ago