Footstep AI

Footstep gives agents the location layer they need to act in the real world.

От сообществаРаботаетБез входаГлобальныйБесплатноТолько чтение

Что умеет

  • Get Directions: Get driving, walking, or cycling directions between locations. Returns distance, duration, turn-by-turn steps, and terrain analytics (elevation, difficulty). Optionally include a natur
  • Optimize Stops: Find the fastest order to visit multiple stops. Returns the optimised route with distance/time savings compared to the original order. Geometry is delivered via the `render` envelope:
  • Compare Routes: Compare 2-4 travel modes between the same locations. Returns a side-by-side comparison of distance, duration, and terrain difficulty with a natural language summary. Geometry is delive

Какие данные видит

Нужен ли аккаунт

Не нужен: сервер работает без входа

Footstep gives agents the location layer they need to act in the real world. Turn place names and addresses into coordinates, plan and optimize multi-stop trips, get turn-by-turn directions, build travel-time isochrones, search for points of interest, snap GPS traces to roads, and predict where someone could reach within a given time (research preview - contact support@footstep.ai for access) . Every response is structured JSON designed for LLM consumption.

Как подключить

Как подключить Footstep AI к Claude

Адрес сервера, который нужно вставить в настройках Claude:

https://footstep.run.tools

  1. Откройте Claude (сайт claude.ai или приложение для компьютера).
  2. Перейдите в Настройки → Коннекторы (в английском интерфейсе: Settings → Connectors).
  3. Нажмите Добавить пользовательский коннектор (Add custom connector).
  4. Вставьте адрес сервера, скопированный ниже, в поле URL удалённого MCP-сервера и нажмите Добавить.
  5. В чате нажмите + → Коннекторы и включите новый коннектор.

Пользовательские коннекторы доступны на тарифах Free, Pro, Max, Team и Enterprise. На Free можно добавить только один. В Team и Enterprise коннектор сначала добавляет владелец организации в Настройки организации → Коннекторы.

Авторизация

После нажатия Добавить откроется окно входа в сервис. Войдите под своим аккаунтом и подтвердите доступ. Пароль Claude не видит.

Как подключить Footstep AI к ChatGPT

Адрес сервера, который нужно вставить в настройках ChatGPT:

https://footstep.run.tools

  1. Откройте ChatGPT в браузере (chatgpt.com). Нужен тариф Plus, Pro, Business, Enterprise или Edu.
  2. Один раз включите режим разработчика: Настройки → Приложения → Дополнительные настройки → Режим разработчика (Settings → Apps → Advanced settings → Developer mode).
  3. Откройте Настройки → Коннекторы и нажмите Создать (Create).
  4. Заполните форму: Название (любое), Описание (одна фраза, что делает сервис), URL MCP-сервера (скопируйте ниже).
  5. В поле Аутентификация выберите OAuth, если сервис требует вход, иначе Нет. Нажмите Создать.
  6. В новом чате откройте + → Приложения/Коннекторы и включите коннектор.

Название раздела у OpenAI менялось (Connectors → Apps/Plugins). Если пункт называется иначе, ищите «режим разработчика» в настройках. В рабочих пространствах Business/Enterprise администратор должен разрешить пользовательские коннекторы.

Авторизация

При первом использовании ChatGPT откроет окно входа в сервис. Войдите и подтвердите доступ.

Как подключить Footstep AI к Cursor

Адрес сервера, который нужно вставить в настройках Cursor:

https://footstep.run.tools

Быстрый способ: нажмите кнопку Открыть в Cursor ниже, Cursor сам предложит добавить сервер.

Вручную:

  1. В Cursor откройте Settings → Cursor Settings → MCP и нажмите Add new global MCP server.
  2. Вставьте JSON, скопированный ниже, в открывшийся файл ~/.cursor/mcp.json (для одного проекта: .cursor/mcp.json в корне проекта).
  3. Сохраните файл. Сервер появится в списке MCP; при необходимости пройдите авторизацию по кнопке рядом с ним.

Авторизация

Если сервис требует вход, рядом с сервером в списке MCP появится кнопка авторизации.

Список инструментов сервера (15)

Технические названия из tools/list. Нужны только разработчикам.

get_directionsGet driving, walking, or cycling directions between locations. Returns distance, duration, turn-by-turn steps, and terrain analytics (elevation, difficulty). Optionally include a natural-language summary. Geometry for the route is delivered via the `render` envelope as a GeoJSON LineString layer.
optimize_stopsFind the fastest order to visit multiple stops. Returns the optimised route with distance/time savings compared to the original order. Geometry is delivered via the `render` envelope: a LineString for the route and a Point layer for the ordered stops (each carrying an `order_index` property).
compare_routesCompare 2-4 travel modes between the same locations. Returns a side-by-side comparison of distance, duration, and terrain difficulty with a natural language summary. Geometry is delivered via the `render` envelope: one LineString layer per mode, each carrying `mode` and `recommended_color` style hints so a renderer can colour them distinctly.
find_and_routeGeocode an origin AND a destination AND compute a route between them, in one call. Use this when the user wants directions between two places given by name — 'directions from A to B', 'route from A to B', 'take me from A to B'. Do NOT use this for 'Where is X?' / 'Find X' — those have no origin and want only the destination plotted; use `geocode` instead. Both `origin` and `destination` accept either lat/lon coordinates OR a place name / postcode / address (which is geocoded internally). Returns the route as a `render` envelope LineString plus Point layers for the matched origin and destination.
get_isochroneFind everywhere reachable within a time or distance limit from a starting point. Use this for 'what's reachable from X in Y minutes' / 'where can I get to from here' queries. Pass `from` as a place name (e.g. 'Kings Cross, London') and the origin is geocoded for you, or pass `location` if you already have coordinates. Returns polygon boundaries for each contour. Geometry is delivered via the `render` envelope as one Polygon layer per contour, with `style_hints.contour_value` carrying the underlying numeric (minutes / kilometres) so renderers can colour by gradient.
get_matrixCalculate travel times and distances between every combination of origins and destinations.
get_elevationGet elevation data for a list of coordinates. Returns heights in metres above sea level and summary statistics (total ascent/descent, min/max elevation, all in metres).
snap_traceMap-match a GPS trace to the road network. Returns the corrected path with confidence score, terrain analytics, and per-segment road attributes (surface, class, speed, grade). Geometry is delivered via the `render` envelope as a GeoJSON LineString of the snapped trace.
search_along_routeFind places along a route ranked by how little detour they add. Useful for finding stops (coffee shops, petrol stations, etc.) on the way. Geometry is delivered via the `render` envelope: a LineString for the route plus a Point layer for the candidate places.
geocodeResolve a single address, landmark, or place name to coordinates. Use this when you specifically need the lat/lng of one place (e.g. to plot it, store it, or feed it to another system). For 'find X near Y' queries, prefer `search_places` with the `near` parameter — it geocodes internally. Returns ranked candidate locations with confidence scores in `results`. The `render` envelope contains ONLY the top match (one point) so a default render answers the question — disambiguation alternatives stay in `results` and can be rendered separately via `addLayer` if the user wants to see them.
reverse_geocodeGiven coordinates (`lat` + `lon`), return the address or place label at that point. Use this when you have a location pin and need a human-readable description (e.g. labelling a map click). For 'what's near here' / 'find X near here' queries, use `search_places` instead.
batch_geocodeGeocode up to 1,000 addresses in one call. Returns coordinates and confidence for each.
search_placesFind points of interest near a location. `query` accepts both categories (cafe, restaurant, supermarket, park, gym, atm, pharmacy, hotel, hospital, fuel, library, museum, …) and specific venue names (Costa, Tesco, Black Sheep Coffee, Tower Bridge). Categories are expanded internally to multilingual synonyms so 'cafe' also catches 'coffee', 'café', 'kaffee', etc. Pass `near` as a place name (e.g. 'Kings Cross, London') and it's geocoded for you, or pass `lat`+`lon` if you already have coordinates. Returns ranked results sorted by distance from the search point.
parse_addressParse and correct messy free-text addresses into structured components. Fixes typos, expands abbreviations (st → Street, ave → Avenue), and infers missing fields like postcode, region, or country. Returns the cleaned address, structured components, an itemised list of corrections, and a confidence score per row. Use this as a pre-step before geocoding when you have dirty input data; pass clean addresses straight to `geocode` or `batch_geocode`.
get_predictionGenerate spatial probability predictions for missing person scenarios. Given a search area, behavioural profile, and environmental conditions, returns per-hex scores across an H3 grid — telling you where to look first. Typically 5–10 seconds for cached areas; large or previously unseen search areas can take up to ~30 seconds while features are computed and cached. Wait for the response; do not retry or assume failure on slow turns. Input hygiene for callers: `profile` is required and SAR-critical — never guess it; ask the operator. `age_years` is optional — omit when unknown; the model handles missing age. `datetime` defaults to the time of the call upstream — pass an ISO 8601 string for any other time. Weather is fetched automatically from `datetime` unless overridden via `weather` or suppressed via `disable_weather: true`.