Paris Traffic: Road Counts & Taxi Stands

Paris Traffic: Road Counts & Taxi Stands MCP Connector for Claude

A+

Keyless Paris road-traffic data: hourly vehicle counts, mean speed and congestion state at ~3,700 permanent counter sites, plus the 420 official taxi stands — filter by site code, state, arrondissement.

7 tools Official Updated Oct 1, 2026 Official Vinkius Partner

The City of Paris permanent road-traffic counters and its official taxi stands — keyless, straight from the Paris open-data platform (opendata.paris.fr).

What you can do

  • list_traffic_sites / traffic_site_lookup / count_traffic_sites — the ~3,700 counter-site reference (code iu_ac, road name, neighboring counters, measurement period)
  • traffic_counts — hourly measurements at one site (vehicle count q, mean speed k, congestion state), with an ISO time window on t_1h and an optional state filter
  • traffic_counts_by_state — how many hourly rows sit in a state ("Fluide", "Pré-saturé", "Saturé", "Inconnu"), city-wide or at one site
  • list_taxi_stands / taxi_stand_lookup / count_taxi_stands — the ~420 official taxi stands: name, arrondissement, address, bay count, status

Who is this for

Anyone reasoning about Paris congestion — which counter is on a given street, how busy it was in a window, how long a corridor stayed saturated — and anyone looking for an official taxi stand. Filters match stored values exactly (the states keep their accents: "Pré-saturé"). A bare time window over the 29.6M-row counter dataset is rejected by the platform, so traffic_counts always reads one site.

parisfrancetrafficcongestiontaxiroad-counterskeyless

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

count_taxi_stands

"75120"), statut ("en service" / "station provisoire") or emplacements — the same exact-match filters as list_taxi_stands. A fast total count; use it to compare stand counts between arrondissements. Count taxi stands, optionally by arrondissement or status

list_traffic_sites

Each row has the site code (iu_ac, e.g. "1232"), the road name with underscores as stored (libelle, e.g. "Av_Gambetta"), the upstream/downstream neighboring counter codes and names, and the measurement period start/end dates. Use iu_ac as the join key for the counting tools. The iu_ac filter is an exact match. Page with limit/offset (100 rows max per call). List the permanent road-traffic counter sites in Paris

taxi_stand_lookup

g. "Porte de Pantin" or "OPEA" (stored "OPERA"). The names are stored without accents in many cases and the match is exact, so use the stored form from list_taxi_stands when unsure. Returns 0, 1 or a few rows with the address, arrondissement INSEE, bay count and status. Find taxi stands by name

traffic_counts

6 million rows, one row per site per hour, with the hourly vehicle count (q, ~42% of rows null), the mean speed in km/h (k), and the congestion state (etat_trafic: "Fluide", "Pré-saturé", "Saturé" or "Inconnu"). iu_ac is required (a bare time window over the whole dataset is rejected by the platform). since/until are ISO timestamps (YYYY-MM-DDTHH:MM:SS, optionally with a UTC offset) delimiting the t_1h column. Pass a few hours, not months — rows are capped at 100 per call, so page with offset to read longer windows. An optional etat_trafic filter is an exact match on one of the four stored states. Read hourly traffic counts at a counter site

traffic_counts_by_state

A fast total count on the 29.6-million-row dataset — use it to gauge how long a corridor stays in a state (e.g. how many hours a site spent "Saturé" recently, combined with the since/until window of traffic_counts). No since/until window is applied here; the count spans the whole dataset. Count hourly rows in a congestion state, optionally at one site

traffic_site_lookup

g. "1232"): road name as stored (underscores), the neighboring upstream/downstream counters and the measurement period dates. iu_ac is required. 0, 1 or a few rows come back (some road names carry several codes). Look up a traffic counter site by its iu_ac code

list_taxi_stands

Each row has the stand name (nom, e.g. "OPERA" for the Opéra stand), the INSEE code of the arrondissement it sits in (insee, "75101".."75120", the 751xx form), the street address, the number of parking bays (emplacements, a number stored as text, "2" up to "20"+) and the status (statut: "en service" for almost all, "station provisoire" for a few). Filters are exact stored values. The whole dataset fits in one or two calls; page with limit/offset (100 rows max). List official taxi stands (bornes d’appel) in Paris

See how to talk to your AI agent using Paris Traffic: Road Counts & Taxi Stands.

Which traffic counter sits on Avenue de Gambetta, and how many cars passed at 8am today?

Look up the site via traffic_site_lookup or list_traffic_sites with libelle "Av_Gambetta", then read traffic_counts for its iu_ac with an ISO window covering 08:00 — the row returns q (vehicles/hour), k (mean speed km/h) and etat_trafic.

How many hours did the counter at iu_ac 1232 stay saturated in the last week?

Call traffic_counts_by_state with iu_ac "1232" and etat_trafic "Saturé" — it returns the number of hourly rows in that state; combine with traffic_counts to list the actual hours.

List the official taxi stands in the 8th arrondissement.

Run list_taxi_stands with insee "75108" (the 8th arrondissement's INSEE code) — each row has the stand name, address and bay count.

No. opendata.paris.fr serves all city datasets anonymously over its Opendatasoft REST API. This MCP defines no credentials and needs nothing configured.

Related Connectors