NYC Open Data Explorer

NYC Open Data Explorer MCP Connector for Claude

A+

Keyless explorer over the 3,000+ datasets of NYC Open Data: catalog search, dataset metadata with column lists, generic SOQL row browsing, grouped stats and row counts — no API key.

5 tools Official Updated Oct 1, 2026 Official Vinkius Partner

The meta-interface to the city of New York's public data. Instead of one fixed dataset, this MCP lets an agent discover and query the whole NYC Open Data platform (data.cityofnewyork.us) — the same Socrata SODA API the city publishes.

What you can do

  • Search the catalog — keyword search over the 3,000+ datasets (id, name, description, update time, monthly views)
  • Inspect a dataset — full metadata: description, last update, download count and the complete column list
  • Browse any dataset — generic row query with SOQL filters, column projection, sort, limit and offset
  • Dataset stats — total row count, optionally filtered, plus top-N values of any column (grouped counts)
  • Row counts — exact totals for sizing a question before pulling rows

Notes

  • All access is anonymous; no credentials are defined.
  • SOQL string literals use double quotes (borough="QUEENS"); date comparisons use ISO "YYYY-MM-DD". Column names must match the dataset exactly — inspect them with get_dataset_info first.
  • The ny-* suite also ships purpose-built MCPs for the highest-value dataset families (311, buildings, traffic, crime, food, TLC, parking, climate, citylife).
new-yorknycopen-datasocratasodacataloggovernment-public-data

5 tools expose this connector's capabilities to your AI agent.

browse_dataset

Filters use SOQL: string values must be wrapped in DOUBLE quotes (borough="QUEENS"), date comparisons use ISO "YYYY-MM-DD" (created_date >= "2026-01-01"), and only double-quoted comparisons — no case-insensitive matching. Check the exact column names with get_dataset_info first. Without parameters it returns the newest-looking sample rows; pass order to control the sort. Query rows of any NYC Open Data dataset with SOQL filters

dataset_row_count

Use it for quick "how big is this dataset" checks; for the breakdown by column use dataset_stats instead. Exact total row count of a dataset, optionally filtered

dataset_stats

group_by must be an exact column name from get_dataset_info. Total row count of a dataset, optionally filtered and grouped by a column

get_dataset_info

Use it before browse_dataset to learn the exact column names. The id is the 8-character value returned by search_datasets (format like "erm2-nwe9"). Get the columns, description and update time of one NYC Open Data dataset by id

search_datasets

Returns the dataset id, name, short description and last update time for each match. Use the id with get_dataset_info (columns and details) or browse_dataset (row queries). Start with short keywords like "potholes", "311", "trees" or "taxi". Search the 3,000+ dataset NYC Open Data catalog by keyword

See how to talk to your AI agent using NYC Open Data Explorer.

Which NYC datasets are about potholes?

search_datasets("potholes") returns the matching datasets (e.g. DOT Pothole Work Orders, 311 Street Condition requests) with ids and last update; browse one with browse_dataset.

What columns does the 311 service requests dataset have?

get_dataset_info("erm2-nwe9") returns the full column list (unique_key, complaint_type, descriptor, borough, incident_address, latitude, longitude, ...), the dataset description and its last row update.

How big is the 311 dataset and what are the top complaint types?

dataset_row_count("erm2-nwe9") gives the total (22M+ rows); dataset_stats with group_by "complaint_type" returns the most frequent complaint types with counts.

No. NYC Open Data serves all public datasets anonymously over its SODA API (data.cityofnewyork.us). This MCP defines no credentials and needs nothing configured.

Related Connectors