Expanding Malware Security Advisories Beyond npm

GitHub has integrated OpenSSF's malicious package data into the GitHub Advisory Database, expanding the scope of malware security advisories and Dependabot alerts from npm to eight major package ecosystems. The supported ecosystems are npm, PyPI, Maven, RubyGems, NuGet, Go, crates.io, and PHP Composer.

Previously, GitHub handled only npm malicious packages detected directly by GitHub through a separate internal pipeline. However, since its launch in 2023, the OpenSSF malicious-packages repository has accumulated over 15,000 OSV-format reports, covering various types such as typosquatting packages, dependency confusion attacks, account takeovers, and malicious pre-built binaries.

Instead of building eight new detection systems for each ecosystem, GitHub created a single repository-based importer. The importer reads files changed since the last run, validates each OSV record against the schema for required fields, data types, and formats, and then updates the database. Records that fail validation are rejected and logged rather than modified to pass.

The collected data is normalized into source, identifier, CVE ID if present, a snapshot of the original record, and mapping information required for the publishing pipeline. This process handles issues such as differing ecosystem names like PyPI and pip, OSV impact scopes recorded per individual version, reports lacking version information or detailed descriptions, and withdrawn advisories.

Duplicate collection is also prevented separately. Since GitHub contributes its own npm malware advisories to the OpenSSF repository, importing them directly would result in re-importing its own data. The importer checks the OSV origin metadata to exclude self-reported records marked as ghsa-malware. During actual verification, more than half of the newly incoming npm reports each month were filtered out using this method.

Malware advisories must be deployed quickly when attacks are ongoing, so they are published automatically without prior human review. With this expansion, automatically published advisories can now generate Dependabot alerts. To support this, GitHub applied validation and deduplication procedures that consider the possibility of upstream data errors leading to security alerts.