health-check-in - SKILL.md Agent Skill

name: health-check-in description: Guided daily or weekly health check-in workflow with adaptive questioning version: 1.0.0

Health Check-In Skill

This skill provides a structured, interactive health check-in workflow that adapts to the user's active conditions and medications. Use this for daily or weekly health monitoring.

Purpose

Guide users through a comprehensive health check-in that covers medication adherence, vital measurements, symptoms, and diet. The check-in adapts based on active conditions and produces a summary at the end.

Prerequisites

Before starting the check-in, gather context:

Get active conditions: condition_list(status="active")
Get active medications: medication_list(active_only=true)
Get recent measurements to understand what vitals the user typically tracks

Check-In Flow

Follow this structured flow in order. Use progressive disclosure — ask one question at a time and wait for response before proceeding.

1. Medication Adherence (if applicable)

If the user has active medications:

For each active medication:

Ask: "Did you take [medication name] [dosage] today?" (or "this week" for weekly check-ins)
If yes: Log the dose using medication_log_dose(medication_id=..., notes=...)
If no: Ask if there was a specific reason and record in notes
If skipped doses: Remind about importance of adherence

2. Vital Measurements

Based on active conditions, prompt for relevant measurements:

Standard vitals for everyone:

Weight (if tracked regularly)
Ask: "Would you like to log your weight today?"
Use: measurement_log(type="weight", value=..., unit="kg")

Condition-specific vitals:

If user has diabetes or "blood sugar" condition:
- Ask: "What's your blood glucose reading?"
- Use: measurement_log(type="blood_glucose", value=..., unit="mg/dL")
If user has hypertension or cardiovascular condition:
- Ask: "What's your blood pressure reading?"
- Accept format like "120/80"
- Use: measurement_log(type="blood_pressure", value={"systolic": 120, "diastolic": 80}, unit="mmHg")
If user has respiratory condition:
- Ask: "What's your oxygen saturation (SpO2)?"
- Use: measurement_log(type="oxygen_saturation", value=..., unit="%")
If user tracks heart rate:
- Ask: "What's your resting heart rate?"
- Use: measurement_log(type="heart_rate", value=..., unit="bpm")

Progressive approach:

Start with most critical measurements based on conditions
Offer to log additional measurements if user volunteers them
Don't overwhelm with too many questions at once

3. Symptom Screening

Screen for symptoms based on active conditions:

General screening:

Ask: "Are you experiencing any new or unusual symptoms today?"
If yes: Ask for symptom name, severity (1-10), and any additional details
Use: symptom_log(name=..., severity=..., notes=...)

Condition-specific screening:

If diabetes: Ask about fatigue, thirst, frequent urination, blurred vision
If cardiovascular: Ask about chest pain, shortness of breath, dizziness, palpitations
If respiratory: Ask about coughing, wheezing, shortness of breath
If gastrointestinal: Ask about nausea, pain, changes in appetite
If chronic pain: Ask about pain location and intensity

For each reported symptom:

Ask severity on 1-10 scale (1=mild, 10=severe)
Ask if symptom is new or ongoing
Log using symptom_log(name=..., severity=..., notes=...)

Important: Only ask about condition-relevant symptoms. If user has no active conditions, do general screening only.

4. Diet and Nutrition (optional)

Ask: "Would you like to log any meals from today?"

If yes:

For each meal: Ask when it was eaten (approximate time is fine), get description, approximate calories if known, and any relevant nutrients
Always pass eaten_at — this is required. Use the user's stated or estimated time. If they say "lunch around noon", use today at 12:00. For planned/future meals, use the expected time.
Use: meal_log(type=..., description=..., eaten_at=..., nutrition={...})

If user is tracking specific dietary concerns (based on conditions):

Diabetes: Ask about carbohydrate intake
Cardiovascular: Ask about sodium intake
Weight management: Suggest using nutrition_summary to review calorie intake

5. Summary Generation

After completing the check-in, generate a summary using this template:

# Health Check-In Summary
**Date:** [current date]
**Type:** [daily/weekly]

## Medication Adherence
[List each medication with adherence status]
- [Medication name]: ✓ Taken / ✗ Missed [reason if provided]

## Vital Signs
[List measurements logged]
- [Type]: [value] [unit] [trend indicator if available]

## Symptoms
[List symptoms reported]
- [Symptom name]: Severity [1-10] [notes]
[Or "No new symptoms reported" if none]

## Diet
[Summary of meals logged or "No meals logged"]

## Notes
[Any additional observations or concerns]

## Follow-Up Actions
[Suggest any recommended actions based on the check-in]
- [Action item 1]
- [Action item 2]

Trend indicators:

Get latest previous measurement using measurement_latest(type=...)
Compare current vs previous value
Show: ↑ (increased), ↓ (decreased), → (stable)

Follow-up recommendations:

If missed multiple medications: Suggest setting reminders
If concerning symptom severity (≥7): Suggest contacting healthcare provider
If vital signs outside normal range: Suggest monitoring or medical consultation
If tracking incomplete: Offer to set up scheduled check-ins

Summary Storage

After presenting the summary to the user:

Offer to save summary as a research note
If accepted: research_save(topic="Health Check-In [date]", content="[summary text]")
This makes check-in summaries searchable later via research_search

Adaptive Tips

For daily check-ins:

Focus on medications and symptoms
Keep vital measurements light (2-3 key metrics)
Quick flow, 5-10 minutes

For weekly check-ins:

More comprehensive measurement review
Use trend_report to show weekly trends
Review medication adherence rate using medication_history
More detailed nutrition review using nutrition_summary

User preferences:

After first check-in, ask user preference for daily vs weekly
Note preference in AGENTS.md for future sessions
Adapt questioning style based on user engagement

Error Handling

If user doesn't know a measurement: Skip it, don't pressure
If user reports concerning symptoms: Prioritize logging those first
If check-in is interrupted: Offer to resume where left off next time
If medication ID lookup fails: Use medication name to search and confirm with user

Example Check-In Flow

Bot: "Good morning! Let's do your daily health check-in. First, let's check on your medications."

Bot: "Did you take your Metformin 500mg this morning?" User: "Yes, just took it." Bot: [Logs dose] "Great! Did you take your Lisinopril 10mg?" User: "Not yet, I'll take it with breakfast." Bot: [Notes for reminder] "Okay, I'll note that. Now let's check your vitals."

Bot: "What's your blood glucose reading this morning?" User: "112 mg/dL" Bot: [Logs measurement] "That's within normal range. Would you like to log your blood pressure today?" User: "Yes, it's 128 over 82." Bot: [Logs measurement] "Got it. Any symptoms you'd like to report today?" User: "No, feeling good." Bot: [Generates summary and offers to save]

Bot: "Here's your check-in summary: [displays formatted summary]. Would you like me to save this summary for future reference?"