Prompt Library

ChatGPT prompts for ADHD planning, focus and weekly resets

30 copy-paste prompts

Copy-ready prompts for task breakdowns, body-doubling scripts, routines and weekly resets. Adults only. Not a diagnosis. Not medical advice.

In short: This page contains 30 copy-paste ready prompts, organized into 6 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Drafted with AI writing assistance and checked against our editorial policy.

Louis Corneloup
By Louis Corneloup · Founder, Techpresso
Last updated ·Drafted with AI assistance (how we make prompts)

Task Breakdowns

5 prompts

One scary task into tiny steps

1/30

✨ What it does

Splits one scary task into eight sub-10-minute steps that use tools you already have, starting with a physical first move. Vague done-states stay tagged; no diagnosis is invented.

Break one scary task into tiny steps I can start today. Task: [SCARY TASK]. What done looks like, in my words: [DONE LOOKS LIKE]. Tools I already have: [TOOLS I HAVE]. Output 8 numbered steps, each doable in under 10 minutes, using only those tools. Step 1 must be a physical first move, such as open the file. If done-looks-like is vague, write [VERIFY] on the last step and ask me to name a file, inbox or room that would count as finished. Do not invent a diagnosis, a therapist homework or a medication. This is a checklist, not medical advice.

💡

Pro tip: Name the file, the sink or the email thread. 'Sort my life' is not a task.

Open loops from a messy list

2/30

✨ What it does

Turns a messy list into open loops with a next physical action, capping tonight at three. Actions that need unlisted tools stay tagged; no extra project is invented.

Sort this messy list into open loops I can close. List: [PASTE MESSY LIST]. Energy I have tonight, in my words: [ENERGY TONIGHT]. Output a table with columns Loop, Next physical action, Minutes if I wrote any, Tonight or later. Tonight gets at most three loops. If a loop has no next action I could do without buying something, write [VERIFY] in Next physical action. Do not invent a project I did not list. Do not label me with a disorder. A made-up 'you probably also need to file taxes' is extra shame, not help.

💡

Pro tip: Paste the notes app dump, including half-sentences.

Start in two minutes version

3/30

✨ What it does

Rewrites one task as a two-minute start that uses something already on your desk, plus a stop rule you chose. An empty desk stays tagged; no system is invented.

Rewrite this task as something I can start in two minutes. Task: [TASK]. Where I am sitting: [WHERE I AM]. What is already on the desk: [DESK ITEMS]. Output a 6 line card: Two-minute start, Then if I keep going, Parking lot for extra ideas, Stop rule I chose in [STOP RULE]. The two-minute start must use an item on the desk. If the desk list is empty, write [VERIFY] and ask me to name one object within reach. Do not invent a productivity system. Do not invent a brain type. A pomodoro I did not ask for is still a hurdle.

💡

Pro tip: If the start needs a login on another device, pick a worse start that is here.

Context-switch cost table

4/30

✨ What it does

Tables each work stream against unfinished next actions and switches you recorded yesterday. Unrecorded switches stay tagged; no ping count or disorder is invented.

Show me the cost of switching using work I already juggle. Work streams: [WORK STREAMS]. Switches I made yesterday: [SWITCHES YESTERDAY]. Output a table with columns Stream, Last unfinished next action from my notes, Switch I recorded, Cost in my words. If I did not record a switch, write [VERIFY] in that cell rather than guessing I flipped ten times. Do not invent a focus disorder. Do not invent Slack pings I did not mention. A story about 'ADHD context switching' I did not write is a diagnosis costume.

💡

Pro tip: Count yesterday from the browser history or the calendar, not from vibes.

What done looks like card

5/30

✨ What it does

Defines done, not-done, who sees it and where the file lives from evidence you already accept. Missing evidence stays tagged as a short pick-list; no boss is invented.

Write a done-looks-like card for one task so I can stop. Task: [TASK]. Evidence I will accept: [EVIDENCE I WILL ACCEPT]. Who else needs to see it, if anyone: [WHO NEEDS TO SEE IT]. Output four lines: Done when, Not done while, Who to send it to, Where I will put the file from [WHERE IT LIVES]. If evidence is missing, write [VERIFY] and list three candidate proofs I must pick from, without adding a fourth. Do not invent a boss, a grade or a clinical outcome. A fake 'inbox zero' I did not define will keep me polishing all night.

💡

Pro tip: Done should be visible, such as 'email sent', not 'I feel ready'.

Body-Doubling Scripts

5 prompts

Twenty-five minute sit-with-me script

6/30

✨ What it does

Writes a 25 minute sit-with-me script with start, mid check and stop lines for a task you named. An unnamed partner stays tagged so the script still works alone; no clinic is invented.

Write a 25 minute body-doubling script I can read aloud or paste to a friend. Friend or group: [WHO IS SITTING WITH ME]. Task I will do: [TASK]. Output a timed script: 0:00 say the task, 0:02 start, 20:00 check whether the next action is still the same, 25:00 stop and name one thing that moved. If I did not name who is sitting with me, write [VERIFY] on 0:00 and keep the script usable alone. Do not invent a coach. Do not invent a clinic group. This is not medical advice and a 'therapy session' I did not book is the wrong frame.

💡

Pro tip: Mute video if faces distract you. The point is another person in the room, not a meeting.

Text a friend to sit with you

7/30

✨ What it does

Drafts a 40 to 70 word text with a time window, a one-line task and an easy no. Missing times stay tagged; no prior agreement or diagnosis is invented.

Draft a short text asking a willing friend to body-double. Friend: [FRIEND NAME]. When I want to start: [START WINDOW]. What I will work on, in one line: [TASK LINE]. Output a text of 40 to 70 words: ask, time window, task line, how they can say no. If the start window is missing, write [VERIFY] where the time would go. Do not invent that they already agreed. Do not mention a diagnosis. Do not claim a medical need. A guilt trip about my brain is not an invitation.

💡

Pro tip: Send it to someone who already said this kind of sit is okay, not to a random group chat.

Virtual coworking intro

8/30

✨ What it does

Writes a six-line coworking intro covering camera rule, your task, midpoint ping and ending. A blank camera rule stays tagged as optional; no attendee names are invented.

Write an intro I can paste at the start of a virtual coworking session. Session: [SESSION NAME]. Camera rule I want: [CAMERA RULE]. Tasks on my list: [TASKS]. Output a 6 line intro: hello, camera rule, my task, how we ping at the midpoint, how we end. If camera rule is blank, write [VERIFY] and default to cameras optional in the intro. Do not invent other attendees' names. Do not invent a paid community. Do not frame this as treatment. A fake 'accountability therapist' in the Zoom is not a coworker.

💡

Pro tip: Put the task in the intro so you cannot silently switch to email.

Mid-session check-in questions

9/30

✨ What it does

Gives five mid-session questions, starting with whether you are still on the named task. Missing elapsed minutes stay tagged in the header; no score or symptom question is added.

Write mid-session check-in questions a body-double can ask me. Task: [TASK]. Minutes elapsed: [MINUTES ELAPSED]. Output exactly five questions, numbered, each yes or no or a next-action in five words. Question 1 must be 'Are you still on [TASK]?'. If minutes elapsed is blank, write [VERIFY] in the header and still ask the five. Do not invent a productivity score. Do not ask about symptoms. Do not diagnose why I drifted. A question about my childhood is not a check-in.

💡

Pro tip: Answer out loud. Typing a novel in the chat is another context switch.

End-of-session recap

10/30

✨ What it does

Creates a recap form for what moved, what is open, tomorrow's first step and where you parked it. Missing park place stays tagged; no win or streak is invented.

Write an end-of-session recap I fill, not a pep talk you invent. Task: [TASK]. What moved, in my words after I type it: I will paste [WHAT MOVED] after we start, so first output a form with blanks What moved, What is still open, First step tomorrow, Where I parked it from [PARK PLACE]. If park place is missing, write [VERIFY] on that blank. Do not invent a win. Do not invent a streak. Do not congratulate me for a brain chemistry I did not describe. A recap with fake progress trains me to lie to the next session.

💡

Pro tip: Fill it before you close the laptop, even if the honest answer is 'I opened the file'.

Daily Routines

5 prompts

Morning launch from my real morning

11/30

✨ What it does

Sequences at most seven steps from your real wake time to the first block, using actions you already do. Missing wake time stays tagged; no cold shower or supplement is invented.

Build a morning launch sequence from what I already do, not from a 5am fantasy. What I already do after waking: [CURRENT MORNING]. Time I actually wake: [WAKE TIME]. First work or care block: [FIRST BLOCK]. Output a sequence of at most 7 steps between wake and first block, using only actions in [CURRENT MORNING] plus one 'put the phone in [PHONE HOME]' if I named a phone home in [PHONE HOME]. If wake time is blank, write [VERIFY] and do not invent an alarm. Do not invent cold showers, journaling I skip or a supplement. Not medical advice: a routine I will not run is decoration.

💡

Pro tip: Time yesterday morning. If it was 40 minutes of phone, write that down.

Shutdown ritual at the actual end of day

12/30

✨ What it does

Writes a six-step shutdown for a stop time you can keep, including device parking and tomorrow's first task. Missing stop time stays tagged; no journal or sleep product is invented.

Write a shutdown ritual for the time I actually stop, not the time I wish I stopped. Stop time I can keep: [STOP TIME]. Devices I must park: [DEVICES]. Tomorrow's first task if I know it: [FIRST TASK TOMORROW]. Output a 6 step shutdown: save, park devices in places I named in [PARK PLACES], write tomorrow's first task, leave the room. If stop time is missing, write [VERIFY] on step 1. Do not invent a gratitude journal. Do not invent a sleep supplement. Do not diagnose insomnia. A shutdown that needs a new hobby will not happen tonight.

💡

Pro tip: Park the laptop charger in another room so opening it again costs a walk.

Leave-the-house object chain

13/30

✨ What it does

Orders a door chain from the furthest object to the exit using homes you named. Homeless objects stay tagged until you pick a hook or bowl; no gadget is invented.

Build an object chain I can run at the door. Objects I must have: [OBJECTS]. Where each one lives: [WHERE THEY LIVE]. Output a numbered chain that starts at the furthest object and ends at the door, each step 'pick up X from Y'. If an object has no home, write [VERIFY] beside it and ask me to pick a hook, bowl or pocket before we continue. Do not invent a gadget. Do not invent a medical alert item I did not list. A key dish I do not own will not save Thursday.

💡

Pro tip: Put a bowl you already own by the door tonight, then rerun the chain.

Meal-time scaffold from my kitchen

14/30

✨ What it does

Builds a five-line meal card from food at home, a slot and minutes you will spend. An empty kitchen stays tagged as cupboard questions; no diet is invented.

Scaffold one meal time from food I already have. Kitchen: [FOODS I HAVE]. Meal slot: [MEAL SLOT]. Minutes I will spend: [MINUTES]. Output a 5 line card: What I will eat from the list, What I will skip, Timer I will set from my minutes, Where I will sit from [WHERE I SIT], What I will not open (phone, extra tabs) if I named it in [DO NOT OPEN]. If the kitchen list is empty, write [VERIFY] and give me three questions to answer while I open the cupboard. Do not invent a diet. Do not invent calories. Not medical advice and not a nutrition plan.

💡

Pro tip: Eat the boring option that is already cooked. Hunting a new recipe is a stall.

Work to home transition

15/30

✨ What it does

Writes a four-step work-to-home transition from your real commute or doorway, open loops and who meets you. A blank commute stays tagged as doorway-only; no family member is invented.

Write a transition between work and home using the commute or doorway I actually have. How I get home: [COMMUTE OR DOORWAY]. What work still pulls at me: [OPEN LOOPS]. Person or pet who meets me, if any: [WHO MEETS ME]. Output a 4 step transition: closing line I say at work from [CLOSING LINE], commute or doorway action, one sentence I can say to who meets me, park-the-work step. If commute is blank, write [VERIFY] and treat it as a doorway only. Do not invent a family member. Do not invent a therapist script. A fake 'how was your day' ritual with nobody there is noise.

💡

Pro tip: The park-the-work step should be physical, like bag on a hook, not a feeling.

Like these prompts? There are full tutorials behind them.

Learn the workflows, not just the prompts. Easy-to-follow tutorials inside AI Academy, and the library keeps growing.

Try AI Academy Free

Focus Blocks

5 prompts

One-block calendar from hours I have

16/30

✨ What it does

Places a single focus block inside hours you already have, with known interruptions parked. Missing hours stay tagged as a request for a start clock; no four-hour fantasy is invented.

Place one focus block on hours I actually have. Hours free: [HOURS FREE]. Task: [TASK]. Interruptions I already know: [KNOWN INTERRUPTIONS]. Output a one-row calendar: Start, End, Task, Interruptions parked where I named in [PARK PLACE]. Start and end must sit inside [HOURS FREE]. If hours free is blank, write [VERIFY] and stop after asking for a start clock time. Do not invent a four-hour deep-work fantasy. Do not invent a focus drug. A calendar brick I cannot sit through is how I ghost the next one.

💡

Pro tip: Start with 25 to 40 minutes you can keep, then book a second block if the first one happens.

Distraction parking lot

17/30

✨ What it does

Maps each usual intrusion to one line you will write in a park place you named. A missing park place stays tagged until you pick paper, file or memo; no extra worry is invented.

Build a parking lot for thoughts that arrive mid-block. Thoughts that usually arrive: [USUAL INTRUSIONS]. Where I will park them: [PARK PLACE]. Output a two-column card: If this thought arrives, Then I write this one line in the park place. Limit the card to the intrusions I listed. If park place is missing, write [VERIFY] and ask me to pick a paper, a note file or a voice memo. Do not invent a thought. Do not invent a rumination diagnosis. A parking lot full of extra worries I did not name is another task list.

💡

Pro tip: Use paper if the phone is the distraction. A notes app on the same phone is a trap.

Phone in another room protocol

18/30

✨ What it does

Sets where the phone goes, what you tell anyone who might need you and how you retrieve it when your timer ends. Missing rooms stay tagged; no lockbox is invented.

Write a protocol for putting the phone in another room during a block. Rooms I have: [ROOMS]. Who else might need me: [WHO MIGHT NEED ME]. Output: where the phone goes, from my rooms; what I tell who might need me, in 20 words; how I get it back when the timer I set in [TIMER MINUTES] ends. If rooms is blank, write [VERIFY] and ask for one other surface besides the desk. Do not invent a lockbox. Do not invent a family emergency plan I did not describe. A protocol that needs a gadget I do not own will fail at 3pm.

💡

Pro tip: Put the phone face down in a room you must walk to. On the sofa arm still counts as here.

If I bounce, restart script

19/30

✨ What it does

Writes a five-line restart: notice, close the bounce destination, stand, sit, two-minute start. A missing start stays tagged as open the file; no shame lecture is added.

Write a restart script for when I bounce off a block. Task: [TASK]. Where I tend to go instead: [WHERE I BOUNCE]. Output 5 lines I can read: Notice I bounced, Close the other tab or room from my bounce list, Stand up, Sit back, Do the two-minute start from [TWO MINUTE START]. If the two-minute start is missing, write [VERIFY] and use 'open the file or paper for [TASK]'. Do not invent a shame lecture. Do not invent a symptom tracker. A restart that takes twelve steps is another bounce.

💡

Pro tip: Practise the stand-up once on a good day so it exists when you need it.

Energy-matched pick from my own words

20/30

✨ What it does

Matches each task to an energy word you used and picks at most one. Missing energy words stay tagged as a request for three friend-words; no clinical label is invented.

Pick today's task from a list using energy words I already used, not a clinical scale. Energy words today: [ENERGY WORDS]. Task list: [TASK LIST]. Output a table with columns Task, Energy word from my list that fits, Why in 12 of my words, Pick or wait. Pick at most one. If energy words are missing, write [VERIFY] and ask me for three words I would tell a friend. Do not invent a hypo/hyper label. Do not invent a diagnosis. Matching a task to 'tired' I did not say is fiction.

💡

Pro tip: Use ordinary words such as foggy, wired, okay. Skip internet ADHD slang if it is not how you talk.

Weekly Resets

5 prompts

Sunday forty-minute reset

21/30

✨ What it does

Times a Sunday reset of 40 minutes or less across areas you named, using next actions you already know. Unknown next actions stay tagged and skipped; no deep clean is invented.

Fill a 40 minute Sunday reset from areas I already want to touch. Areas: [AREAS]. Minutes I will give: [MINUTES], which must be 40 or less. Output a timed list that adds up to my minutes, with each area getting a next action I could do in that slice. If an area has no next action in [NEXT ACTIONS I KNOW], write [VERIFY] and skip it rather than inventing a deep clean. Do not invent a meal prep I do not cook. Do not invent a therapy homework. A 3 hour 'reset' is how Sundays disappear.

💡

Pro tip: Set a kitchen timer. When it rings, the reset is over even if the sink is still there.

Calendar versus dumped tasks

22/30

✨ What it does

Marks each dumped item as already on the calendar or not, then park or schedule. Items without a duration stay tagged; no meeting or deadline is invented.

Reconcile my calendar with a dumped task list. Calendar: [PASTE CALENDAR]. Dump: [PASTE DUMP]. Output a table with columns Dump item, Already on calendar yes or no, Next action, Park or schedule. If a dump item has no duration I wrote, write [VERIFY] in Next action rather than inventing a 2 hour block. Do not invent a meeting. Do not invent a deadline. A phantom Friday review I never run should not appear as if it were booked.

💡

Pro tip: Paste the real calendar export, including the boring appointments you will keep.

What slipped without the shame story

23/30

✨ What it does

Tables slipped items against facts you pasted, with drop or retry and a tiny next step. Unknown instead-facts stay tagged; no moral or disorder reason is invented.

Review what slipped last week using facts I paste, without a character story. Slipped items: [SLIPPED ITEMS]. What I did instead, if I know: [WHAT I DID INSTEAD]. Output a table with columns Item, Fact from my paste, Next tiny step, Drop or retry. If I do not know what I did instead, write [VERIFY] in that cell. Do not invent a moral. Do not invent a disorder reason. Do not write 'because of ADHD' unless those words are in my paste. A shame novel is not a plan.

💡

Pro tip: If you drop an item, delete it from the list so it stops haunting Monday.

Three rocks for next week

24/30

✨ What it does

Chooses exactly three rocks from commitments you already made, each with a done line and a day you can work. Hours that cannot cover three stay tagged on rock 3; no fourth rock is invented.

Pick three rocks for next week from work I already committed to. Commitments: [COMMITMENTS]. Hours I can honestly give: [HOURS]. Output exactly three rocks, each with a done-looks-like line from my words and a day from [DAYS I CAN WORK]. If hours cannot cover three, write [VERIFY] on rock 3 and recommend dropping it. Do not invent a fourth rock. Do not invent a OKR. Do not invent a client. Three honest rocks beat a vision board I will not open.

💡

Pro tip: A rock is visible to someone else, such as a sent draft, not 'think about the project'.

Admin hour contents

25/30

✨ What it does

Maps one protected hour across papers, messages and money admin you already named, with a five-minute buffer. Items needing unnamed logins stay tagged; no bill is invented.

Fill one admin hour with tasks I already owe. Owed admin: [ADMIN LIST]. Hour I can protect: [HOUR WINDOW]. Output a minute map that fits the window, grouped as papers, messages, money admin I named, and a 5 minute buffer. If an item needs a password or a person I did not name, write [VERIFY] and park it. Do not invent a bill. Do not invent a tax form. Do not invent a bank. An admin hour stuffed with extra adulting I did not list is how I avoid the hour.

💡

Pro tip: Put the hour on the calendar with a start ritual, such as headphones on, then start the first paper.

Go from copy-pasting to actually mastering AI.

AI Academy: hands-on tutorials on ChatGPT, Claude, Midjourney, and many other tools. New tutorials are added regularly.

Start Your Free Trial

Capture and Reminders

5 prompts

Inbox-ish capture from notes

26/30

✨ What it does

Turns pasted notes into short loops parked in a capture place you already use. Feelings with no action stay tagged and skipped; no second inbox is invented.

Capture open loops from notes I paste into a list I can trust. Notes: [PASTE NOTES]. Capture place I already use: [CAPTURE PLACE]. Output a bullet list of loops, each 8 words or fewer, plus a line 'parked in [CAPTURE PLACE]'. If a note is a feeling with no action, write [VERIFY] and skip it. Do not invent a second inbox. Do not invent a GTD ritual I do not run. A new app in the output is another place to forget.

💡

Pro tip: One capture place. If you have four, pick the one you opened yesterday.

Recurring reminder list from my own items

27/30

✨ What it does

Writes one reminder line per repeating chore in the tool you already use, at cadences you wrote. Missing cadence stays tagged; no pill schedule is invented.

Turn chores I already repeat into reminder lines I can paste into a tool I already use. Chores: [CHORES]. Tool: [TOOL I USE]. Cadence I already try: [CADENCE]. Output one reminder line per chore in the tool's plain language, with the cadence I wrote. If cadence is missing for a chore, write [VERIFY] beside it. Do not invent a weekly reset I skip. Do not invent a pill reminder. Not medical advice: a dose schedule I did not supply is out of scope.

💡

Pro tip: Attach reminders to events that already happen, such as after the kettle, not to a blank hour.

Waiting-on tracker

28/30

✨ What it does

Tables people you already asked, the ask, the date if you have it and a nudge on a channel you use. Missing dates stay tagged; no extra follow-up is invented.

Build a waiting-on tracker from people I already asked. People and what I asked: [WAITING ON]. Date I asked, if I have it: [DATES ASKED]. Output a table with columns Person, Ask, Date or [VERIFY], Nudge date I will choose in [NUDGE RULE], Channel I already use from [CHANNEL]. If a person has no ask, drop the row. Do not invent a follow-up I did not send. Do not invent a manager. A tracker full of ghosts makes the real waits invisible.

💡

Pro tip: If you cannot name the channel you used, you probably did not actually ask.

Visual cues in the rooms I have

29/30

✨ What it does

Places cues you are willing to put up into rooms you live in, each tied to a task. Cues that need unnamed walls stay tagged; no smart light is invented.

Place visual cues in rooms I actually live in. Rooms: [ROOMS]. Cues I am willing to put up: [CUES I WILL PUT UP]. Tasks those cues should trigger: [TASKS]. Output a table with columns Room, Cue from my list, Task it triggers. If a cue needs a wall I did not name, write [VERIFY] and skip it. Do not invent a smart light. Do not invent a housemate's permission. A sticky note in a room I do not have is not a cue.

💡

Pro tip: Use objects you already see, like a basket on the shoe you trip over.

I forgot again recovery card

30/30

✨ What it does

Builds a five-line recovery card from what you forgot, where you found it last time and who you had to tell. Missing last-time locations stay tagged; no punishment is invented.

Write a recovery card for the next time I forget something I already forgot before. What I forgot: [WHAT I FORGOT]. Where I found it last time: [WHERE I FOUND IT]. Who I had to tell, if anyone: [WHO I TOLD]. Output a 5 line card: What to check first, Where to look from last time, Who to message in 15 words, How to add one cue from [CUE I WILL ADD], How to stop the spiral in one sentence I supply in [STOP SPIRAL SENTENCE]. If last-time location is missing, write [VERIFY] on Where to look. Do not invent a diagnosis. Do not invent a punishment. Not medical advice: a forgetting is a logistics problem here, not a label.

💡

Pro tip: Keep the card in the same place as the thing you forget, even if that feels silly.

Free tool

Prompt Optimizer

Turn a rough idea into a structured, professional AI prompt.

Try it free →

Frequently Asked Questions

No. They are planning drafts for adults who already want task breakdowns, body-doubling scripts, routines and weekly resets. This is not medical advice and not a diagnosis. Care, medication and therapy stay with clinicians you already see.
Adults only. They assume you can choose your own hours, rooms and friends. They are not for diagnosing a child, writing school plans or substituting for a clinician.
It will if you let it. These prompts ask for your real wake time, rooms and tools, and they mark gaps [VERIFY]. Delete any cold shower, new app or 5am block you did not write.
Paste them to a person who already agreed to sit with you, or read them alone with a timer. Keep the task in the first line. Skip anything that sounds like a clinic group you did not join.
A calendar export, a dumped task list and hours you can honestly give. Pick three rocks you can show someone. Drop slipped items you will not retry so they stop haunting Monday.

Prompts are the starting line. Tutorials are the finish.

A growing library of hands-on tutorials on ChatGPT, Claude, Midjourney, and many other AI tools. New tutorials are added regularly.

Start 7-Day Free TrialCancel anytime.
Explore AI for Consultants