Appfigures

What Appfigures MCP server does

The Appfigures MCP server exposes Appfigures app analytics and market-intelligence operations as tools for MCP-compatible clients. It runs through the af mcp CLI command and communicates over stdio, so an MCP client starts the process locally and sends tool requests to it.

The available command groups cover app discovery, catalog exploration, numeric metrics, store data, audience information, and reviews. App search returns unified app records and store listings, while app lookup can show metadata and the data available for tracked apps. Explorer operations work across millions of products on Apple, Google Play, Amazon, and other major stores, with fields covering identity, availability, categories, ratings, ranks, estimates, SDKs, demographics, and related apps.

How it works

Install or invoke the npm package, authenticate with Appfigures, and register af mcp in the MCP client configuration. The server uses the CLI’s authenticated session and makes the corresponding Appfigures operations available to an agent. A globally installed CLI can be launched as af; clients can instead run the package with npx -y @appfigures/cli mcp.

Metrics queries can target numeric datasets for one or more apps and can group results by up to two dimensions. Country, device type, date range, and app-selection filters can narrow a query. Explorer queries support filtering and sorting, while aggregate queries return counts, averages, minimums, maximums, or histograms rather than individual product records.

Store tools provide rank histories, category charts, category IDs, featured placements, and localized listing data. Audience tools cover estimated age and gender distributions and cross-usage. Review tools support reading and working with iOS and Google Play reviews, including filters such as rating, date, country, version, and tracking relationship.

Setup and configuration

The Appfigures MCP server requires Node.js 22 or newer. Run npm install -g @appfigures/cli for a global installation, or use npx without installing the package. Authenticate interactively with af auth login --interactive, or complete the authorization URL and code flow with af auth login followed by af auth login --code <code>.

Authentication normally saves the token in the operating system credential manager: macOS Keychain, Windows Credential Manager, or Linux Secret Service. For unattended or headless use, set APPFIGURES_API_KEY to a personal access token created from the Appfigures developer keys page.

A basic client entry uses command npx and arguments -y, @appfigures/cli, and mcp. Add APPFIGURES_API_KEY under the client’s env object when the process should not depend on stored credentials. Claude Code can register the server with claude mcp add appfigures -- npx -y @appfigures/cli mcp.

Tools and capabilities

The Appfigures MCP server supports these functional areas:

• Find, inspect, and list tracked apps.
• Search and aggregate the broader app catalog.
• Discover catalog fields and available numeric datasets.
• Query metrics across apps, dates, countries, and device types.
• Inspect store ranks, charts, categories, featured placements, and listings.
• Read audience demographics and app cross-usage.
• Search, filter, summarize, and reply to app store reviews.

Limitations and notes

Some operations depend on the user’s Appfigures account and access level. Metrics queries without an app filter cover the apps tracked by the account. Store listing requests operate on one storefront product and one locale at a time. App identities can be unified across storefronts, but store-specific operations may require the numeric product ID for a single storefront.

Explorer and metrics results are structured for analysis rather than general app records: explorer aggregation returns summary statistics, and metrics returns a nested partition tree. The CLI documentation identifies available datasets and catalog-field access levels, so agents may need to inspect those descriptions before constructing a query.