Seoul Essentials

Seoul Essentials — MCP Server

Essential public facility data for AI agents helping foreign tourists in Seoul, South Korea.

What is this?

Seoul Essentials is an MCP (Model Context Protocol) server that provides structured data about essential public facilities in Seoul. Designed for AI agents that assist foreign tourists with real-time queries about nearby services.

Available Data (37,900+ places, 17 types)

Counts are the live Firestore document counts as of 2026-08-04.

MCP Tools

find_places

Search for public facilities by type and conditions.

Parameters:

• type (required): one of the 17 types listed above (e.g., "toilet", "pharmacy", "post_office", "traditional_market", "ev_charger")
• district (optional): Seoul district name in English or Korean (e.g., "gangnam", "강남구")
• filters (optional): Service-specific filters (e.g., {"english": true}, {"is_24h": true}, {"indoor": true})
• limit (optional): Max results, 1-50 (default: 10)

get_place_detail

Get full details of a specific place by its ID.

Parameters:

• id (required): Place ID (e.g., "toilet_00001", "pharmacy_001", "aed_00001")

find_nearby

Find public facilities near GPS coordinates, sorted by distance.

Parameters:

• lat (required): Latitude (Seoul range: ~37.4 to ~37.7)
• lng (required): Longitude (Seoul range: ~126.7 to ~127.2)
• radius_m (optional): Search radius in meters, 100-5000 (default: 500)
• type (optional): Filter by facility type
• limit (optional): Max results, 1-20 (default: 5)

get_subway_timetable

Get subway timetable for a specific station.

Parameters:

• station (required): Station name in Korean or English (e.g., "강남" or "Gangnam")
• line (optional): Line number (e.g., "2")
• day_type (optional): "weekday" | "saturday" | "holiday" (default: "weekday")
• direction (optional): "up" | "down"

Quick Start

Option A: Smithery (Recommended)

Visit Seoul Essentials on Smithery and follow the setup instructions for your MCP client.

Option B: Direct Connection

Add to your MCP client config (e.g., Claude Desktop claude_desktop_config.json):

{
  "mcpServers": {
    "seoul-essentials": {
      "url": "https://seoul-essentials-mcp-230987091625.asia-northeast3.run.app/mcp"
    }
  }
}

Option C: Run Locally

git clone https://github.com/do-droid/seoul-essentials.git
cd seoul-essentials
API_BASE_URL="https://asia-northeast3-seoul-essentials.cloudfunctions.net/api" \
uv run python -m src.server

Example Queries

An AI agent can use these tools to answer questions like:

English

• "Find an English-speaking pharmacy near Gangnam Station"
• "Where is the nearest public restroom to my location?"
• "What's the subway schedule at Myeongdong Station?"
• "Find free WiFi hotspots in Hongdae area"
• "Where is the closest AED to Itaewon?"

日本語

• 「明洞で日本語が通じる薬局を探して」
• 「弘大の近くで無料Wi-Fiが使える場所は？」
• 「江南駅周辺のEV充電スポットを教えて」

中文

• "在明洞附近找会说中文的药店"
• "弘大附近哪里有24小时便利的公共厕所？"
• "梨泰院附近的清真餐厅在哪里？"

Feedback

We actively review and incorporate feedback to improve data quality and coverage.

For AI agents: Use the submit_feedback tool during a conversation to report issues or suggest improvements.

For developers: Open a feedback issue on GitHub.

Categories: new_data | data_quality | coverage | format | frequency | other

Known Data Limitations

• pharmacy: Two layers. 417 records come from Seoul Open Data Plaza's "Foreign-language Capable Pharmacy Status" survey and carry services.english / chinese / japanese; Gangnam-gu is absent from that upstream survey entirely (origin gap, not a parser bug), which is why ?filter.english=true&district=gangnam still returns nothing. The remaining 5,455 come from the MOIS LOCALDATA licence register and cover all 25 districts with precise coordinates and phone numbers. Licence-sourced records carry services.language_support: "not surveyed" and deliberately omit the language keys — they are never advertised as English-capable, and ?filter.english=true excludes them.
• post_office / traditional_market / ev_charger: Upstream datasets do not include lat/lng. Coordinates fall back to the centroid of each Seoul district (coordinates_approximate: true). District-level filters (?district=mapo) are exact; find_nearby ranking against these types is district-grained rather than building-grained.
• toilet: Opening hours come from a free-text field with 234 distinct formats. 2,008 records are confirmed 24-hour; 445 publish no usable hours and carry an explanatory hours.note instead of open/close times.
• Opening hours in general: is_24h: true appears only where the source supports it — toilets, 따릉이 docks, outdoor WiFi, EV chargers and tourist zones. Every other type's source publishes no hours at all, so those records carry is_24h: false plus a hours.note saying what is actually known. Filtering ?filter.is_24h=true therefore returns only verified round-the-clock places.
• address_en: 71.9% of records carry an English road address generated by Revised Romanization (scripts/romanize.py in the backend repo). Building number, road and district are exact; a few loanword road names romanize phonetically rather than to their official spelling (올림픽로 → "Ollimpik-ro", officially "Olympic-ro"). Lot-number (지번) addresses carry no road name and are left empty rather than guessed.
• Roadmap (next round): halal_friendly (data.go.kr 15111159 was withdrawn — searching alternative sources) and emergency_room (data.go.kr 3043449 not findable — switching to e-gen.or.kr OpenAPI) are planned for the following refresh. Both were removed from the tool schema in 0.4.0 — they had been advertised as valid types while holding zero documents, so every call returned an empty result.

Data Sources

All data sourced from official Seoul government open data:

• Seoul Open Data Plaza (data.seoul.go.kr)
• Korea Open Data Portal (data.go.kr)

License

MIT