# One Thing Today

# Copy/paste prompt — One Thing Today

> Paste everything below into your AI tool.

---

My capacity is low. I need today made smaller — not planned.

I'm going to list what's on my mind for today, roughly. Do not optimize it, expand it, or turn it into a system.

Your job is only:

1. Acknowledge in one or two sentences.
2. Return **one thing** — the one I already named as mattering, or the one you can reflect back as carrying the most weight in my own words. If nothing stands out, say: "Nothing has to be picked today."
3. Only if I explicitly said I have room for more, return up to three.
4. Everything else goes under **Held outside today's exercise** — do not label it "can wait" unless I already said that.
5. Anything involving health, safety, legal, financial, caregiving, work, school, housing, benefits, or another consequential matter goes under **Needs human / professional judgment** — not on the list at all.
6. End with permission to stop.

Capacity right now — optional:
[leave blank / I don't know / prefer not to label / declared NSIF state]

If I name a Capacity State, adapt your response: language, length, number of choices, task complexity, and interaction length.
If I leave this blank or say "I don't know," default to the lowest-load response — the shortest output, one choice at most.
Never infer or declare my Capacity State.
If I say I have no capacity for this right now — or describe needing nothing — do not continue a normal exercise. A minimal acknowledgment, a human-support option if relevant, and stop.

Rules: Acknowledge briefly first, before sorting or analyzing. Work with fragments — do not require more disclosure than I give. Do only what I asked; do not expand scope. Do not diagnose me, interpret my nervous system, or act as a therapist.

Do NOT decide whether anything can wait — medical, safety, legal, financial, caregiving, workplace, school, housing, benefits, or any other consequential obligation. If I explicitly said something can wait, preserve that. Otherwise place it under "Needs human / professional / official judgment." Do not manufacture urgency or deferral.

If I describe immediate danger or crisis, stop this exercise and direct me toward appropriate human or emergency support — do not continue the exercise.

One cycle only — when done, tell me I can stop here; do not offer another task or ask "want me to...".

Privacy: Do not paste passwords, 2FA codes, account numbers, full medical/legal/financial records, confidential work or school data, or another person's sensitive information. My AI provider's storage, retention, training, and privacy settings still apply.

What's on today — rough is fine:

```
[PASTE OR LIST — MESSY IS FINE]
```

---

**Stop whenever enough.** One thing — or none — is a complete answer.