visit-prep-pack-pdf

name	visit-prep-pack-pdf
description	Build the downloadable visit-prep pack PDF for the day's clinic — cover with practitioner + date + roster, then one full pre-visit-summary page per patient. Renders via the inline HTML template + Playwright.
enabled	true

Instructions

Use this skill when Dr. Solomon asks for "the pack", "the PDF", "print my prep", or any variation. This is the deliverable the persona exists to produce.

Workflow

1. Gather the data by chaining other skills first:
   • The roster: daily-schedule flow against PRA-9001 for the requested date (default to today, 2026-06-02)
   • Per-patient context: for each booked patient, in parallel:
     • epic_get_patient
     • epic_list_conditions
     • epic_list_medications (status: active)
     • epic_list_allergies
     • epic_list_observations (limit: 10 recent)
     • The current epic_get_encounter for the day's appointment (for reason_text if present)
2. Render by invoking scripts/render_visit_prep.py via code_interpreter. Pass the data as JSON:

python /app/use-cases/clinician-visit-prep/skills/visit-prep-pack-pdf/scripts/render_visit_prep.py \
  --practitioner-name "Dr. Aniyah Solomon" \
  --practitioner-specialty "Internal Medicine, Adult Primary Care" \
  --clinic-date 2026-06-02 \
  --patients-json '<JSON array of patient records>' \
  --out /tmp/visit-prep-2026-06-02.pdf

3. Confirm in chat: "Visit-prep pack saved to /tmp/visit-prep-<YYYY-MM-DD>.pdf — N pages: cover + one per patient. Allergies in red, out-of-range labs flagged." Reference the path so file-sharing picks it up.

Patients-JSON shape

The script expects:

[
  {
    "patient": { "id": "PAT-100001", "mrn": "MRN-100001", "first_name": "Eleanor", "last_name": "Hsu", "date_of_birth": "1948-03-12", "sex": "F", "insurer": "...", "address_city": "Boston", "address_state": "MA" },
    "encounter": { "id": "ENC-200003", "start": "2026-06-02T10:00:00Z", "type": "Office Visit", "reason_text": "Follow-up: BP not at goal", "duration_minutes": 30 },
    "conditions": [{ "code": "I10", "display": "Essential hypertension", "onset": "2014", "status": "active" }],
    "medications": [{ "name": "Lisinopril 20 mg PO daily", "indication": "Hypertension", "since": "2014" }],
    "allergies": [{ "substance": "Penicillin", "severity": "moderate", "reaction": "rash" }],
    "observations": [{ "code": "BP", "value": "148/92", "unit": "mmHg", "effective_at": "2026-04-15", "interpretation": "above goal" }],
    "focus_notes": ["BP control — 148/92 above target", "LDL above post-MI goal"]
  },
  ...
]

focus_notes is your suggested-focus bullet list — synthesise these from the data + the clinical-guidelines-reference skill.

Assets

• assets/visit-prep-pack.html — the Jinja-style HTML template (no Jinja, just {{PLACEHOLDER}} string substitution). Clinical white-and-teal theme; cover page; one page per patient.

Constraints

• One page per patient (use page-break-before CSS); cover stays on page 1.
• Render severe (🔴) allergies above the problem list. Moderate (⚠️) inline.
• File size budget < 2 MB. Don't embed images other than the Olympus mark in the template.
• Do not include real PHI — fixture data only.
• The "Suggested focus" bullets should reference guideline section numbers when relevant (e.g. "LDL 82 — above §3.1 target <70 post-MI").

When NOT to use

• Single-patient brief — use pre-visit-summary and answer inline.
• Roster only (no per-patient depth) — use daily-schedule and answer inline.

Available Files

This skill directory contains the following files you can read with read_file using their absolute paths (prefix /app/use-cases/clinician-visit-prep/skills/visit-prep-pack-pdf/):

• assets/visit-prep-pack.html
• scripts/render_visit_prep.py