No preamble. Here are five prompts, why each one works, and exactly when to use it.
Prompt 1: The Anchor Finder
When to use: Before you start any new habit. This is the first session.
I want to build the habit of: [describe target habit]
My typical weekday schedule (be specific about recurring events):
Morning: [describe]
Midday: [describe]
Afternoon/evening: [describe]
My typical weekend schedule: [describe briefly]
I need you to find me one reliable anchor behavior I can attach this new habit to. An anchor should: (1) happen at roughly the same time, in roughly the same context, at least 5 days a week; (2) be something I already do automatically, not aspirationally; (3) occur close in time to when I'd ideally do the new habit.
Ask me clarifying questions about my routine until you have enough to identify the best anchor. Then give me the completed anchor sentence: "After I [anchor], I will [new habit]."Why it works: Most habit attempts fail at the cue. Vague cues (“in the morning”) are unreliable. This prompt forces a specific, stable anchor and uses the conversation to find it in your actual life rather than an idealized version.
Prompt 2: The Behavior Sizer
When to use: Once you have an anchor. Run this before your first day.
My target habit: [what you ultimately want to do]
My anchor cue: [from Prompt 1]
My current situation: [beginner / returning to this habit / have some consistency but not automatic]
Honest time available daily: [realistic number — not aspirational]
Please design three versions of this behavior:
1. Starter behavior — takes under 2 minutes, requires almost no motivation, something I'd do even on the worst Wednesday of the year
2. Intermediate behavior — what this looks like after the starter is automatic (roughly 30 days)
3. Target behavior — the full habit I'm working toward
For each version: tell me exactly what "done" looks like (not vague like "I exercised" — concrete like "I did 10 push-ups after my coffee").
Then ask me: "Does the starter behavior still feel too large?" and adjust if I say yes.Why it works: B.J. Fogg’s core insight is that starter behaviors need to be below the motivation threshold — so small you’d do them when tired and distracted. This prompt operationalizes that principle and builds in a check against the most common mistake (starter steps that are still too large).
Prompt 3: The Identity Statement Builder
When to use: After Prompts 1 and 2. Run once at the start; revisit at day 30.
I'm building the habit of: [describe habit]
I'm at the beginning — I've done it: [0 / a handful of times / inconsistently for a while]
The kind of person I want to become: [describe in your own words — not ideal self, actual direction]
What doing this habit consistently would mean about me: [your honest answer]
Write me three identity statements I could say or write immediately after completing this habit:
1. A modest version that's completely true right now
2. A directional version — a slight stretch, aspirational but not self-deceptive
3. An ambitious version for 6 months from now
Important: these should sound like how I'd actually talk to myself, not affirmations I'd roll my eyes at. If any version feels fake, say so and adjust.Why it works: Identity-based habits are more durable than outcome-based ones, per James Clear’s synthesis of the behavioral literature. But the mechanism only works if the identity language feels genuine. This prompt builds in the authenticity check — the explicit instruction to flag anything that feels fake.
Prompt 4: The Weekly Review
When to use: Every week, same day, same time. Five minutes. Start using this from day one.
Weekly habit review — [today's date]
Habit: [name]
Days completed this week: [X] out of [Y available]
Missed days: [list them, one sentence each on what was happening]
How automatic does this feel right now (1-10, where 10 = I do it without thinking): [number]
One thing I've been avoiding noticing or admitting: [something you're aware of but haven't said directly]
Please:
1. Identify the most likely structural reason for any missed days — design problem, timing problem, motivation problem, identity friction, or competing demands. Be specific.
2. Suggest exactly one change I should make this week. Not "try harder" — a specific adjustment to the anchor, the behavior design, or the timing.
3. Tell me: maintain, scale up, or simplify?
4. Give me one question to sit with before next week's review.Why it works: This is the prompt that makes everything else work. The weekly review is the feedback loop that converts a habit attempt into a habit-building practice. The “one thing I’ve been avoiding” line consistently surfaces the most useful diagnostic information — the thing you know but haven’t examined.
Prompt 5: The Failure Diagnosis
When to use: Any time a habit has stalled for more than a week, or when you’re thinking about abandoning it.
My habit has stalled. Here's the situation:
Habit: [describe]
How long I've been attempting it: [weeks/months]
Consistency over the last two weeks: [X days out of Y]
The reason I've been telling myself for missing days: [the official story]
What was actually happening on missed days: [the more honest version, if different]
Number of previous attempts at this same habit: [number]
What pattern I see in previous failures: [if any]
Please:
1. Ask me three questions that might surface something I've been avoiding looking at.
2. Diagnose the primary failure mode: design failure, context failure, motivation failure, identity friction, or capacity failure. Explain your reasoning.
3. Give me two options: (a) how I could modify this habit to make it actually work, and (b) what it would look like to consciously put this habit on pause rather than failing at it indefinitely.
Don't just encourage me to keep going. Tell me what you actually see.Why it works: The explicit instruction to give an honest assessment matters — AI models default toward encouragement. This prompt overrides that by giving explicit permission to deliver hard feedback. The three-way diagnosis (design / motivation / identity) surfaces the right solution, because the solution is different for each category.
For the full system these prompts fit into, see the step-by-step guide to building habits with AI. For the framework behind the diagnostic categories in Prompt 5, see the HABIT Loop framework.
Your action: Run Prompt 1 right now. Open Claude or ChatGPT, paste it with one habit you’ve been meaning to build, and describe your daily routine. You’ll have an anchor cue in ten minutes.
Tags: AI prompts for habits, habit building prompts, AI habit coaching, behavior change prompts
Frequently Asked Questions
-
Which AI should I use for these habit prompts?
Both Claude and ChatGPT work well. Claude tends to produce more nuanced, contextual responses for the identity and failure diagnosis prompts — it handles the qualitative, reflective parts better when you give it detailed context. ChatGPT is strong for the design prompts and structured output. Either works. The prompt quality matters more than the model.
-
Can I use these prompts more than once?
Yes — and you should. The weekly review prompt (Prompt 4) is designed to be used every week. The failure diagnosis prompt (Prompt 5) should be run any time a habit stalls. The other prompts can be revisited if you redesign the habit. Save the prompts somewhere accessible — you'll use them repeatedly.
-
How specific should I be when filling in the brackets?
As specific as possible. The more context you give — your actual daily schedule, your specific failure history, your real constraints — the more useful the output. Generic inputs produce generic outputs. Five sentences of genuine context outperforms a single vague sentence every time.