dadhalfdev/skill-google-maps-reviews

This repo is a small Python skill that scrapes reviews from Google Maps for a given place and returns them in a clean JSON format. The user only has to supply a search term, such as a restaurant name and city, and the script finds the matching location on Google Maps, pulls up to 300 reviews from it, and optionally translates each one into a chosen target language, with English as the default. The intended caller is an AI agent or automation script that wants real customer feedback as input data. Each review in the output keeps both the original text and the translated version, so downstream code can choose which one to work with. Dates that Google shows as relative phrases like "3 weeks ago" are converted into proper ISO timestamps so the reviews can be sorted by time. The output also includes the star rating from 1 to 5, the detected language of the review, and a small profile for the reviewer with their display name, profile link, total review count, profile picture URL, and Local Guide level when Google exposes it. The README lists several example use cases: tracking a business's reputation by pulling recent reviews and flagging complaints, comparing customer sentiment across competing restaurants or hotels, doing quick market research before opening in a neighbourhood, feeding the text into a sentiment or topic classifier, populating a retrieval based question answering system, or wiring the data into reports, dashboards, and CRM workflows. The skill runs headlessly by default, meaning it controls a hidden browser without opening a window, and there is a headed mode for local debugging. It is designed to slot into agent setups like Cursor, LangChain, CrewAI, and MCP based tools, but it is just a Python script and can be called directly from any other code that knows how to invoke it. For users who would rather not run the scraper locally, the author also points at a hosted version on Apify under the same name, which adds scheduling, cloud storage, CSV and Excel export, and MCP tool access while accepting the same core input fields: search_term, max_reviews, sort_by, and target_language. The repo itself currently has zero stars and is presented as one of several agent ready scraping skills from the same author.