Reference¶
Data Model¶
The ossiq domain model is located in the ossiq.domain module. It defines the core entities used for analysis.
Project¶
A software project being analyzed. Each Project contains a name and lists of its direct production and development dependencies.
For full details, see ossiq/domain/project.py.
Package¶
A dependency of a Project. A Package is defined by its name and contains a list of all its available versions.
For full details, see ossiq/domain/package.py.
Version Models¶
The version-related models capture details from different sources and are aggregated into a single Version object.
The primary Version object aggregates package_data (from a package registry) and repository_data (from a source code repository). Other data classes like Commit and User provide granular detail about the source code history.
For a complete definition of all version-related data classes, see ossiq/domain/version.py.
System Behavior¶
Dependency Resolution¶
Dependency Graph: The system operates on a flat list of dependencies resolved from a lockfile (e.g.,
package-lock.json).Transitive Dependencies: Transitive dependency resolution is not performed. The tool relies on the dependency resolution of the target project’s native package manager (e.g.,
npm,pip,uv).
Data Provenance¶
Package metadata is sourced from ecosystem-specific repositories (e.g., npm registry, PyPI). This is handled by a set of adapters in the ossiq.adapters module (e.g., ossiq.adapters.api_npm).
Analysis Output¶
A single analysis run produces a ProjectMetrics object.
Class: ossiq.service.project.ScanResult
Description: Contains an analysis of each dependency, including version lags, time lags, and associated vulnerabilities.
Data Sources¶
OSS IQ aggregates data from the following public sources:
Source |
Purpose |
|---|---|
Open-source vulnerability database (CVEs, security advisories) |
|
License and curation data for open-source packages |
|
Package metadata and version history for JavaScript packages |
|
Package metadata and version history for Python packages |
|
Repository activity, releases, and maintainer signals |
Outputs¶
OSS IQ produces three categories of analysis — metrics, security, and supply chain exposure — delivered across four output formats.
Metrics¶
Each dependency produces a PackageMetrics record with the following measurements:
Metric |
Field |
Description |
|---|---|---|
Version lag |
|
Days elapsed since |
Release lag |
|
Releases between |
Drift status |
— |
Semantic classification: MAJOR, MINOR, PATCH, LATEST, or NO_DIFF |
Metrics Operationalization¶
time_lag_days and releases_lag are deterministic numbers. Teams use them to define thresholds that match their risk tolerance and enforce them automatically in CI.
A typical starting point:
Threshold |
Field |
Recommended starting value |
|---|---|---|
Maximum version age |
|
365 days |
Maximum release distance |
|
— (use |
Use the JSON export to evaluate thresholds in a CI step:
# Fail if any production package is more than 365 days behind
MAX_LAG_DAYS=365
jq --argjson max "$MAX_LAG_DAYS" \
'[.production_packages[] | select(.time_lag_days != null and .time_lag_days > $max)] | length' \
ossiq-report.json
Start with a permissive threshold to baseline your project, then tighten it incrementally as tech debt is resolved. This avoids blocking CI on day one while still creating measurable improvement targets.
For a complete GitHub Actions setup with CVE gating and outdated-package blocking, see the Version Lag and CVE Quality Gate tutorial.
Security¶
Each PackageMetrics record contains a cve array. Each entry includes:
Field |
Description |
|---|---|
|
Primary vulnerability identifier (CVE, GHSA, or OSV ID) |
|
All aliases for this vulnerability (CVE, GHSA, OSV IDs) |
|
Database that reported the vulnerability |
|
LOW, MEDIUM, HIGH, or CRITICAL |
|
Description of the vulnerability |
|
List of affected version strings |
|
Publication date (ISO 8601, nullable) |
|
URL to the upstream advisory |
Transitive CVEs. When a transitive dependency has CVEs, OSS IQ surfaces them in the transitive_packages array. The dependency_path field on each entry traces the ancestor chain from the project root to the affected package.
Supply Chain Exposure¶
OSS IQ identifies two version constraint risk patterns using the version_constraint field:
Risk |
Condition |
Signal |
|---|---|---|
Pinned version |
|
Prevents automatic updates |
Upper-bound constraint |
|
Actively excludes newer versions |
Output Formats¶
Console — Scan Table¶
The scan command prints two dependency tables — one for production packages, one for development — to the terminal. Each row shows package_name, CVE count, drift status, installed_version, latest_version, releases_lag, and time_lag_days.
Console — Package Detail¶
The package command prints a six-section report for a single dependency:
Drift status — version comparison with an ASCII time-lag progress bar
Dependency tree trace — ancestor path from the project root to the package
Policy compliance —
version_constraint, resolved version, and latest versionSecurity advisories — per-CVE
severity,id, source, andsummaryVia transitive dependencies — CVEs inherited from entries in
transitive_packagesLicenses — SPDX identifiers from the
licensefield
HTML Report¶
The scan --presentation html command produces a self-contained HTML file embedding an interactive Vue.js single-page application. The report includes the full dependency tables and the Transitive Dependency Explorer: an interactive D3 tree that visualises the transitive_packages dependency graph.
The Explorer supports:
Color-coded nodes by risk type (CVE, pinned version, upper-bound constraint)
Fuzzy search and toggle filters (CVE, Pinned, UBC)
Click to focus a node and highlight all ancestor and descendant paths
Alt+Click to collapse or expand a subtree
Dashed curved links between nodes sharing an identical
package_name@installed_versionZoom and pan
For full Explorer interaction details, see EXPLORER.md.
JSON Export¶
The export --output-format json command writes a single .json file conforming to export schema v1.1. The root object contains:
Key |
Contents |
|---|---|
|
|
|
|
|
Aggregate counts: packages, CVEs, outdated |
|
Array of |
|
Array of |
|
Array of |
CSV Export¶
The export --output-format csv command writes a folder named export_{project_name}/ containing three files and a Frictionless Data descriptor:
File |
Contents |
|---|---|
|
One row of project metadata and aggregate counts |
|
One row per package with all |
|
One row per CVE with all |
|
Schema references and foreign key relationships |
Versioning & Stability Guarantees¶
OSS IQ makes four commitments to users who depend on its output in CI pipelines, scripts, or downstream tooling.
Export Schema Stability¶
Each export schema version is identified by schema_version in the metadata block (e.g. "1.1"). The export --schema-version flag pins output to a specific version.
Within a schema version:
Existing fields are never renamed or removed.
New optional fields may be added — existing consumers are unaffected.
When a schema version is deprecated, the previous version remains fully supported for at least one major release cycle. Deprecation is announced in the changelog before the version is removed.
CLI Interface Stability¶
Command names, flag names, and exit codes are considered stable interfaces. Changes follow the same deprecation policy as schema versions: the old form continues to work with a deprecation warning before it is removed.
Deterministic Analysis¶
Given the same lockfile and the same version of OSS IQ, a scan always produces the same output. This makes OSS IQ safe to run as a blocking CI gate and suitable for diffing results between runs.
Note
Package registries and source code providers may remove versions or repositories at any time. OSS IQ cannot control this. Scan results may differ between runs if upstream data changes.
Note
Risk scores are time-dependent by design. The same lockfile analyzed at different points in time will produce different scores. A CVE’s risk weight increases the longer it remains unpatched (survival analysis). A new library with high release activity signals different risk than an established library with a stable, slow release cycle — and that signal shifts as the library matures.
Metric Deprecation¶
When a field or metric is deprecated, it continues to appear in exports with its original semantics until the next major schema version. Removal is always accompanied by a migration note describing the replacement field or approach.