mvp-list-propose

name	mvp-list-propose
description	Use when selecting and ranking MVP List places from a Markdown repo to propose trip plans. Applies filters such as location, open hours, infant friendliness, visited/liked history, place type, route fit, and user preferences, then composes three candidate plans.

MVP List Propose

Propose reads schema-shaped Place Markdown records and creates three candidate plans. It does not render final timed agendas; use mvp-list-render after the user picks or approves a plan.

Inputs

• Markdown repo path, usually ./demo-md-repo, ./.tmp/gbrain, or $GBRAIN_REPO.
• Trip date, start/end time, home/base location, driving radius, must-visit places, avoid list, meal needs, infant constraints, visited/liked history.
• If any critical constraint is missing, make a conservative assumption and list it.

Workflow

1. Load places

   • Read *.md files.
   • Parse YAML frontmatter between ---.
   • Validate against ../_shared/references/place.schema.json when practical.
   • Ignore records that cannot be parsed unless the user asks for repair.

2. Filter

   • Type balance: usually one or two point_of_attraction stops plus lunch/dinner restaurants.
   • Location: keep places within the requested area or route corridor.
   • Open hours: remove places clearly closed during intended visit windows; keep uncertain hours with a risk note.
   • Infant friendliness: prefer short walks, parking, picnic/rest areas, flexible timing, easy exits, lower noise.
   • Been-to status: use family_context.visited and liked_if_visited; avoid repeats unless liked or requested.
   • Weather/season/access risks: consider outdoor exposure, closures, weekend-only or weekday-only restrictions.

3. Score

   • Assign a transparent 1-5 score for route fit, hours fit, infant fit, novelty, confidence, and meal fit.
   • Penalize missing critical data; do not hide uncertainty.

4. Compose three plans

   • Plan A: safest/default family-friendly plan.
   • Plan B: scenic or highlight-heavy plan.
   • Plan C: fallback or lower-effort plan.
   • Each plan should include ordered stops, rough windows, why it works, risks, and backup swaps.

Output

## Plan A — <short name>

- Stops: <ordered places>
- Why this works: <route/hours/infant rationale>
- Risks: <closures, missing hours, weather, access>
- Backup swaps: <place -> place>
- Score: <summary>

## Plan B — <short name>
...

## Plan C — <short name>
...

Ask the user to choose one plan or request a merge before using mvp-list-render.