Review Miner

What the Review Miner MCP server does

Review Miner MCP server helps developers investigate what users like and dislike about competing mobile apps and PC games. It works with public App Store and Steam data and returns structured findings rather than requiring an agent to inspect thousands of reviews manually.

The server can identify products in a category, retrieve an app or game ID, summarize recent reviews for one product, and compare multiple competitors. Review analysis includes rating distribution, ratings by version where available, recurring complaint phrases, and selected helpful review text. Comparisons can include two to eight products and can mix App Store and Steam entries.

This fits the Search & Data Extraction category and is useful when researching product gaps, investigating a rating decline, or reviewing competitor feedback across a market or country.

How it works

An MCP client sends tool calls to the local server. The server obtains App Store information through Apple iTunes Search and RSS sources, and Steam information through the Steam Store and Web API. It performs fetching, counting, and phrase extraction before returning a compact result for the model to interpret.

The server limits returned review text to the most helpful samples, which avoids filling the model context with every fetched review. Each tool accepts a country such as tr, de, or us, and results can be requested as Markdown or JSON. Steam review retrieval also supports language filtering, including English, Turkish, or all available languages.

Setup and configuration

Review Miner MCP server requires uv and does not require API keys. With Claude Code, it can be added using claude mcp add review-miner -- uvx review-miner-mcp. Claude Desktop and Cursor can run it through an MCP configuration with uvx as the command and review-miner-mcp as its argument.

The repository also documents local development commands using uv sync --extra dev, offline tests with mocked APIs, a live Apple and Steam check, and the MCP Inspector. These development commands are optional for normal installation.

Tools and capabilities

• review_top_apps retrieves current App Store charts by category and chart type, or Steam weekly lists for top sellers, most played games, and new releases.
• review_search_apps searches for an app or game and returns its identifier.
• review_fetch analyzes recent reviews for one product, including rating mix, rating by version, complaint phrases, and helpful review text.
• review_compare_apps compares two to eight App Store or Steam competitors side by side.
• find_opportunities supports category-based product-gap research.
• competitor_teardown supports a focused love-versus-hate analysis for one app.

The Review Miner MCP server can return tool results in Markdown or JSON. Its analysis is based on source data and computed summaries; the model remains responsible for interpreting the findings and turning them into product conclusions.

Limitations and notes

App Store access is limited to roughly the most recent 500 reviews per country because of Apple's feed limits. Apple's review feed can also temporarily return no results for a particular app and country. The server retries equivalent feed URLs and reports the issue when all attempts remain empty; trying again later or selecting another country may help.

Steam results cover recent reviews and can be filtered by language. Results are cached for 15 minutes, and concurrency is capped to reduce load on the source services. Review Miner MCP server therefore supports recent-feedback analysis, not an unrestricted historical archive.