mvp-list-enrich

name	mvp-list-enrich
description	Use when enriching an MVP List Place Markdown record using all available public data sources. Completes schema-shaped YAML frontmatter for structured search, including restaurant top_dishes and attraction points of interest, then validates against place.schema.json.

MVP List Enrich

Enrich takes a candidate Place Markdown record or a resolved place and fills the structured YAML frontmatter for search. The Markdown body is only for human review; structured fields must live in frontmatter.

References

• Schema: ../_shared/references/place.schema.json
• Markdown template: ../_shared/references/place-markdown-template.md
• Empty JSON field reference: ../_shared/references/place-template.json

Source Strategy

• Verify dynamic facts from current public sources: official site, Google Maps, Yelp, Apple Maps, OpenTable, delivery apps, booking pages, government/park pages, official PDFs, maps, brochures, trail organizations, and reputable local guides.
• Preserve source URL and retrieval date for every factual group.
• Never fabricate uncertain facts. Use null and confidence_notes.

Required Enrichment

Common fields:

• site_ids: at least one canonical reference.
• location: address, city, region, country, coordinates when source-confirmed.
• open_hours: raw hours, timezone, source URL, retrieval date.
• family_context: visited/liked fields if known; infant/baby notes from evidence or explicit itinerary context.
• sources and confidence_notes.

For restaurant:

• Fill restaurant.cuisine_type, ratings, price_range, and vibe where available.
• Fill restaurant.top_dishes with sourced popular dishes, signature dishes, or high-frequency menu items.
• Dish source priority: official menu, ordering platform, reservation platform, review platform, menu aggregator.
• Each dish must have name, description, and source_url; use description: null if only the name is confirmed.

For point_of_attraction:

• Fill point_of_attraction.category, estimated_visit_duration, and attractions.
• Fill point_of_attraction.attractions with sourced key points of interest: trails, viewpoints, exhibits, buildings, gardens, picnic areas, photo spots, landmarks, or activity zones.
• Attraction source priority: official venue/park page, official map/PDF/brochure, public agency page, conservation/trail organization, reputable local guide.
• Each attraction must have name, description, and source_url.

Markdown Writing

When writing to a directory-style Markdown repo:

1. Resolve target path from user input, $GBRAIN_REPO, ./demo-md-repo, or ./.tmp/gbrain.
2. Use a slug file name like <place-name>-<city>.md.
3. Put the full schema-shaped object in YAML frontmatter.
4. Put readable notes in Summary, Visit Facts, Restaurant Notes or Point Of Attraction Notes, Sources, and Confidence Notes.
5. Parse the --- frontmatter and validate it against ../_shared/references/place.schema.json.
6. If no gbrain query tool exists, report only “Markdown written and schema-valid”; do not claim query indexing.

Output

Updated MVP List Place:

- Place: <name>
- Type: <restaurant | point_of_attraction>
- File: <path>
- Structured status: frontmatter passed schema validation
- gbrain status: <query-verified | written to Markdown, query indexing not verified>
- Main gaps: <none or field list>