android-corpus-prep

What problem does it solve?

Manually gathering, filtering, and tracking the provenance of Android APKs for analysis is time-consuming, error-prone, and lacks reproducibility, making it difficult to build consistent, well-documented test corpora.

Core Features & Use Cases

• Metadata aggregation and fast querying: Fetches AndroZoo and Google Play metadata, converts large gzipped sources to compressed Parquet for low-memory, high-speed DuckDB queries.
• Popular APK selection: Filters for high-quality, popular apps based on download count, star rating, and malware detection status to build targeted corpora.
• Bulk download with provenance: Downloads APKs from Google Play or AndroZoo with per-pass manifests that track source, SHA256 hash, version, selection reason, and file path for full auditability.
• Use Case: A mobile security researcher building a corpus of popular financial apps can filter the pre-built popular APK candidate set for wallet-related packages, download the selected APKs, and generate a complete manifest to feed into downstream decompilation and vulnerability scanning workflows.

Quick Start

Use the android-corpus-prep skill to build a curated corpus of popular financial Android APKs with full provenance tracking for security analysis.