Skip to main content
Software represents a product that may be affected by a vulnerability. It includes packages, applications, operating systems, appliances, hardware devices, and firmware. A Software record is versionless: it identifies the product, not every release of that product. Registry-observed releases are child records linked to Software, while affected version ranges, fixed versions, and remediation remain claims attached to vulnerabilities and advisories.

When to use software

  • Enrich an SBOM, application inventory, operating-system inventory, or device record with related vulnerabilities and advisories.
  • Normalize package identity across ecosystems such as npm, Maven, PyPI, Go, or Debian.
  • Connect affected-version claims from multiple publishers to one exact software identity.
  • Inspect releases observed in the package registry and their vulnerability or advisory assessments.
  • Build software-centered security activity views for product and dependency owners.
Do not interpret every vulnerability related to software as affecting every version. Match the deployed version against affected assertions and inspect provenance before creating remediation work.

Software kinds

Every software record has a kind: Package-specific fields such as purl, ecosystem, namespace, and package_name are optional for Software as a whole. They are populated only when they apply to a package-shaped record.

Identity rules

Zero resolves software only from exact identifiers. It prefers a versionless PURL for packages, then a product-level CPE identity built from part, vendor, and product, then an exact publisher namespace and source product identifier. Display-name similarity alone never merges software records. PURLs preserve package ecosystem, namespace, and name semantics. Version-specific CPE coordinates remain attached to the product as source evidence and affected-version data rather than creating duplicate Software records. A source-scoped identifier keeps a publisher’s product identity distinct when no standard coordinate is available.

Releases

For supported npm, PyPI, Cargo, and NuGet packages, Zero observes release versions and publication metadata from the package registry. Each release remains a child of its versionless Software identity. Release assessments compare the observed version with normalized affected ranges and report affected, unaffected, or unknown together with a reason. General web search does not create release records. Registry observations do not replace publisher affected-version claims. The assessment connects those two records so consumers can distinguish a published release from a claim about whether that release is affected. Assessments are aggregated by vulnerability or advisory: source_count records how many source versions support the result, and conflict plus conflicting_sources makes disagreement explicit. The software_releases projection returns at most 100 releases in ecosystem and ecosystem-correct version order. Each release returns at most 100 vulnerability or advisory subject assessments. Both levels include _meta.total, _meta.limit, and _meta.truncated; consumers must check truncated before treating either list as complete.

Relationships

Schema

Every software detail response starts with these fields: Request optional projections when you need more detail: Search responses are intentionally more compact than detail responses. Use the API pages for the complete request and response schemas.

API pages