Atmy MCP server

• Registry nameai.atmy/air-qualityverified in the MCP registry

The registry name sits under atmy.ai. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.

• Stateanswerschecked 2026-10-28
• Tools11
• Server calls itselfatmy
• Server version1.4.0
• Protocol version2025-06-18

On 2026-10-28 the server answered initialize and tools/list without a sign-in. First answer in 625 ms.

• Its tools do Check current PM2.5 levels in a city, but not Download air quality readings as CSV files.
• Checked on 1 of the 7 days it takes to judge reliability.

Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.

• INITE MCP scoreafter a weeka score needs 7 daily checks: 1 so far
• What an agent can do with it1 of 2 use cases
• Use385 npm downloads a week
• Answeredmeasuring1 of the 7 daily checks it takes to print a share
• Median first answer639 ms
• Space in an agent's context≈ 2,489 tokensthe definitions of all its tools (11), which an agent carries on every turn
• Tool definitionsA · 91 of 100
• Typed results0 of 11 toolsdeclared outputSchema: results come typed, read once

• Parameters without a description: in 6 of 11

A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (25%), doing the product's job (25%), definitions (20%), context (10%), use (10%) and how readily a client connects (10%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →

• search_cities

  Find the cities with an Atmy air-quality summary by name (any language) or country. Returns slugs to pass to the other tools.

• get_air_quality

  PM2.5 now in a city (mean of the latest reading of each eligible sensor, at most 60 minutes old) with its US EPA category and the weather now (temperature, humidity, pressure, wind, clouds, precipitation), plus the last completed day from…

• get_air_quality_history

  Daily PM2.5 of a city from the Atmy Data Hub (UTC days, sensors after the quality check) with the day's weather from 2026-01-19 (mean, lowest and highest temperature, humidity, pressure, wind, strongest gust, clouds): the last days, or…

• get_air_quality_monthly

  The whole measurement history of a city from the Atmy Data Hub: PM2.5 per month with the month's weather from 2026-01, yearly means (complete years marked), the dirtiest and cleanest months and days. For questions like "how has the air…

• get_city_forecast

  The next hours of a city: temperature, wind, humidity, precipitation and the pollution accumulation index 0–100 (how strongly the weather keeps particles near the ground). A weather estimate, not a PM2.5 forecast.

• get_country_ranking

  Cities of a country ranked by daily PM2.5 for the last completed day (Atmy Data Hub), most polluted or cleanest first.

• list_sensor_networks

  The sensor networks in Atmy (Sensor.Community, AirGradient, Atmy) with their number of sensors, how many pass the quality check, and their license.

• find_sensors

  Single sensors with their latest reading (PM1, PM2.5, PM10, temperature, humidity, pressure), network and quality check: status, rating 0–10 of the last checked day, the mean rating of the last 7 days and the reasons. Filter by city…

• get_sensor

  One sensor by its id: location, network, license, quality check with rating and reasons, and the latest reading.

• get_sensor_readings

  Readings of one sensor: raw (the last 31 days, at most 7 days a call), 5-minute means (the same windows), hourly means (the last 366 days, at most 92 days a call) or daily means (UTC days of the last 366 days, with the quality check of…

• get_sensors_average

  The mean of up to 100 sensors per 5 minutes, hour or UTC day, each sensor counted once per bucket, with the number of sensors in each bucket. Windows as in get_sensor_readings; by day at most 92 days a call.