The home health aide arrives Monday morning to fill in for two days while you visit your sister. You hand her three printed sheets — the medication list, the meal preferences, the morning routine — and realize halfway through the explanation that you forgot to mention the new pill that started last week, the napping pattern that's been off, and the fact that the cat's medication is also on the schedule now. You text her later in the afternoon to fill in the gaps. She handles it gracefully. You feel the small, familiar drag of having missed something that mattered.

A care plan is supposed to make this easier. In practice, most care plans live as a Word document that gets updated twice a year and as a hundred small details only you know. The handoff to a sibling, an aide, a respite caregiver, or a hospital intake nurse is always a little improvised — and every improvisation has a small cost.

A vault that holds the care plan as a living document, with the agent quietly assembling whatever handoff you need, makes the gap between *what only you know* and *what someone else can step into* a little smaller. (For the day-to-day appointment side, [Caregiver Notes: Medications, Appointments, and the Care Plan in One Place](/guides/personal-life/caregivers-medications-appointments/) is the companion guide.)

## A care plan is one page, plus the documents and the daily reality underneath

In a markdown vault like Docapybara, the care plan itself can be a single overview page on the person you're caring for — the version someone could read in five minutes and have a working picture of. Below it, pages nest with no depth limit, so the underlying detail can live on child pages: *Medications*, *Daily routine*, *Diet*, *Mobility*, *Cognitive*, *Doctors*, *Emergency plan*.

The overview page links to the detailed pages. When the daily routine changes, you update the routine page and the agent can refresh the overview to match: *"Update the care plan overview to reflect the changes I made on the daily routine page this week."*

For caregivers managing more than one person, each gets a parent page with their own care plan underneath. The structure stays the same; the content reflects each person's reality.

## Medications, with the schedule and the why

A care plan's medication section needs to do two things at once: serve as a quick-reference list (which a doctor or pharmacy or aide can read) and capture the *why* behind each one (which only you know).

An inline `:::database:::` directive on the *Medications* page handles the list — name, dose, time given, prescribing doctor, purpose, started, and notes (e.g. *give with food*, *patient prefers crushed in applesauce*). Six column types are available, so you can mix categorical fields like *time of day* with prose fields like *notes* without forcing everything into one shape.

The agent updates it. *"Add the new beta blocker — name, 25mg morning, started today after the cardiology visit, for blood pressure, give with breakfast."* Row appears.

For the schedule itself — what the day looks like in terms of medications — ask the agent to draft a clean version: *"From the medications database, generate the daily medication schedule, sorted by time of day, formatted for the home health aide who's covering this weekend."* You get a clean handoff document.

The *why* matters because medications get questioned. Why is she on this antidepressant? Why is the diuretic at noon instead of morning? When a new doctor reviews the list — and there will be new doctors — the answers should be on the page so you don't have to recall the reasoning from a conversation eight months ago.

## A daily routine that survives a real handoff

Beyond medications, the most asked-about part of a care plan is the daily routine. What happens when. What the person likes and doesn't. The small habits that feel invisible to you but matter to someone stepping in.

A *Daily routine* page in prose, organized by part of the day, works better than a rigid template. Morning: *up around 7:30, hates to be rushed, prefers tea before the morning meds, breakfast around 8:30, watches the news on channel 5 while eating.* Mid-day: *short walk if weather permits, lunch at noon, naps from about 1 to 2:30.* Evening: *dinner at 6, reading or TV, in bed by 10. Doesn't like the bedroom light on after 9.*

This kind of detail is what makes the difference between an aide who feels like a stranger in the house and one who feels like a help. Voice is a good way to draft it — talk through what a normal day looks like, the small preferences and rhythms, and the transcript becomes the first version of the page. Edit later if needed.

When you're handing off, ask the agent: *"Write a one-page daily routine summary for the aide who's covering this weekend, focused on what she needs to know to make Mom comfortable."* The agent pulls from the routine page and produces a clean handoff. You can read it before you send it.

## Diet, allergies, and the food that actually works

Diet is harder than it looks because it changes — appetites shift, allergies emerge, foods go from acceptable to refused for reasons that aren't always clear. A *Diet* page can hold the current preferences, the things that don't work, the things to absolutely avoid, and the meals that have been hits.

For complex diets — diabetes, dysphagia (swallowing problems), kidney disease, allergies — the page can hold the dietitian's notes (drop the PDFs; they convert to markdown automatically) plus your day-to-day observations. *"Pureed diet started last month after the swallow study. Likes the chicken soup blended thin, won't tolerate the same thing for breakfast two days in a row."*

For meal planning, the agent helps: *"Look at the diet page and the daily routine. Suggest five lunches that fit her preferences and the dietitian's restrictions, that the aide could prep without much effort."* You get a workable list grounded in what actually works in your house. (For the recipes themselves, [Your Recipe Box: Stop Losing the Good Ones](/guides/personal-life/ai-recipe-box-never-lose-recipe/) is the right companion.)

For grocery lists, the same shape: *"Based on the meal plan for next week, draft a grocery list."*

## Mobility, transfers, and the equipment in the house

For people with mobility limitations, the care plan needs to capture how the person moves through their home — what they can do unassisted, what needs help, what equipment is in which room, what's safe and what isn't.

A *Mobility* page in prose, sometimes with a small inline database for the equipment inventory. *Walker in the living room. Wheelchair in the bedroom for longer distances. Shower chair in the bathroom. Grab bars on both sides of the toilet, only on the left side of the shower.*

Transfers — getting from bed to chair, from chair to standing — are where most aides have specific questions. *"Transfers from the bed: she can do it with one-person assist if you brace her right knee. From the recliner: needs to push up with both arms first; don't pull her up by the hands."*

The agent can pull this for a handoff. *"Draft a mobility and transfer summary for the new aide who starts Monday — one page, focused on what she needs to do safely on day one."*

For PT or OT homework — exercises the therapist has prescribed — drop the printout on the page. The agent reads it and can remind you what the routine is supposed to be: *"What's the leg-strength routine the PT prescribed last month, and how often is it supposed to happen?"*

## Cognitive notes, mood, and the patterns that take time to see

For people with cognitive changes — dementia, post-stroke recovery, slow age-related decline — a care plan needs to capture both the current state and the trajectory. What's normal for this person right now. What's a red flag.

A *Cognitive* page can hold both. The current state in prose: *"Knows family by name. Sometimes confuses dates by a few days. Can manage simple tasks (making toast) but can't manage complex sequences (cooking a meal). Repeats the same question two or three times an hour, more in the late afternoon."* Updated as things change.

The trajectory is what the *Daily log* gives you, accumulated over months. The agent can pull patterns: *"Look at the daily log over the last six months. Has the late-afternoon confusion gotten more frequent or more severe?"*

For the aide or family member stepping in, knowing the patterns is what prevents small interactions from going wrong. *"She often asks the same question several times in a row in the late afternoon — answer each time as if it's the first; don't tell her she already asked. She gets agitated by being corrected."* That's the kind of detail that comes from months of paying attention and that nobody else has access to until you write it down.

## Emergency plan, contacts, and the page you hope nobody opens

The care plan needs an *Emergency* page. What to do if she falls, if she's confused beyond normal, if there's chest pain, if she can't be roused. Who to call in what order. Where the legal documents are. The hospital you'd want her taken to.

This page should be short, scannable, and current. The agent can keep it in sync with the doctors page and the legal documents: *"Update the emergency plan to include the new neurologist's after-hours number and the updated POA on file."*

The contact list — every doctor, the home health agency, the pharmacy, the in-network hospital, the family decision-makers — sits on or near this page. When something happens at 9pm on a Sunday, the value of having one number search away is much larger than the effort of building the list.

For caregivers who travel or can't always be reached, the emergency plan should also identify the secondary decision-maker. *"If you can't reach me, call my brother first; if not him, call my cousin who lives in the area."* Aides need this; siblings need it. (Healthcare workers using the same shape professionally see [AI Notes for Healthcare: Patient Context That Survives Handoffs](/guides/field-service-ops/ai-notes-healthcare/).)

## Sharing the plan without sharing the vault

Docapybara is single-user — the vault is one person's. For sharing the care plan with siblings, aides, or a care manager, the right shape is to export clean summaries when you need to.

The agent drafts shareable versions. *"Draft a one-page care plan summary I can email to my brother before his visit next weekend."* Or: *"Write a handoff packet for the new home health aide — daily routine, current medications, mobility notes, dietary preferences, doctors, and what to do in an emergency. Single document, formatted clean."*

The vault holds the source of truth — the long-running notes, the daily log, the changing details — and you generate the handoffs as needed. That keeps the working space yours and the shareable artifacts current.

## A starter shape that works on day one

If you're building a care plan from scratch today, here's a first version:

- **[Person's name]** — one parent page.
- **Care plan overview** — one child page, written in prose, the version someone could read in five minutes.
- **Medications** — one page with the inline database.
- **Daily routine** — one page in prose.
- **Doctors** — one page with contacts.
- **Emergency plan** — one page, scannable.

That's the working set. Pages for diet, mobility, cognitive, hospital stays, and legal documents get added as they become relevant.

For caregivers managing the role part-time around a job — which is most of them — the cost of a vault is mostly the upfront capture. Once it exists, the agent does the assembly work. The next handoff doesn't require you to remember everything; it requires you to ask for the right document.

Caregiving doesn't get easier with better notes. What it does is take a small but real chunk of the cognitive load off your shoulders — and gives the people stepping in a fair shot at doing the job well.

[Try Docapybara free](/accounts/signup/). Start with the medication list and the daily routine, and see how much smoother the next handoff feels.