AppInsightMCP is a Model Context Protocol (MCP) server written in JavaScript that wraps two existing scraper libraries (@jeromyfu/app-store-scraper and @jeromyfu/google-play-scraper) and exposes their functionality as MCP tools. It runs as a stdio-based server that AI clients like Claude Desktop can connect to in order to query mobile app marketplace data.

The problem it addresses is that large language models have no built-in way to fetch live data about mobile apps — details, rankings, reviews, ratings, similar apps, privacy info, permissions, and so on. Rather than each user writing custom scraping code, this server provides a ready-made bridge so an AI assistant can answer questions about apps in the Apple App Store and Google Play Store on demand.

The intended users are developers, product managers, marketers, or researchers who want to use an AI assistant (via MCP) to perform app market intelligence — competitor research, review analysis, category exploration, and trend tracking. Users install it via npx, Docker, or Smithery and configure it in their MCP client's config file.

Internally, the implementation is straightforward: a single src/server.js file registers roughly 20 tools (10 for App Store, 10 for Google Play) using the @modelcontextprotocol/sdk. Each tool defines a Zod schema for parameters, calls the corresponding scraper library function, and returns the JSON result as text. Communication uses StdioServerTransport, so it's driven by the parent MCP client process. A Dockerfile provides a production build using node:22-alpine.

No obvious red flags. The code is a thin, transparent wrapper over public scraper libraries; there is no telemetry, no network calls beyond the scrapers themselves, and no credential handling. Users should be aware that the underlying libraries scrape public store pages, which may be subject to rate limiting or terms-of-service considerations from Apple/Google.