# # Audit policies # audit: alerts: # # category: malicious packages (publicly known and unknown) # malicious: contains known malware: - reason: package is known to contain a dangerous malware - enabled: true typo-squatting or repo-jacking package: - reason: package impersonates another popular package to propagate malware - enabled: true # # alert category: suspicious packages (potentially malicious) # suspicious: inconsistent with repo source: - reason: package code inconsistent with the public repo source code - enabled: false # WIP overwrites system binaries: - reason: package code inconsistent with the public repo source code - enabled: false # WIP # # alert category: packages vulnerable to code exploits # vulnerable: contains known vulnerabilities: - reason: known vulnerabilities (CVEs) in package code could be exploited - enabled: true insecure network communication: - reason: package code uses insecure network communication (not https) - enabled: false # WIP # # packages with undesirable or "risky" attributes # undesirable: package is old or abandoned: - reason: old or abandoned packages receive no security updates and are risky - enabled: true invalid or no author email: - reason: a package with lack of or invalid author email suggests 2FA not enabled - enabled: true invalid or no homepage: - reason: a package with no or invalid homepage may not be preferable - enabled: false no source repo: - reason: lack of public source repo may suggest malicious intention - enabled: true fewer downloads: - reason: a package with few downloads may not be preferable - enabled: true no or insufficient readme: - reason: a package with lack of documentation may not be preferable - enabled: false fewer versions or releases: - reason: few versions suggest unstable or inactive project - enabled: true too many dependencies: - reason: too many dependencies increase attack surface - enabled: false version release after a long gap: - reason: a release after a long time may indicate account hijacking - enabled: false contains custom installation hooks: - reason: custom installation hooks may download or execute malicious code - enabled: false # WIP # # type: repo stats # few source repo stars: - reason: a package with few repo stars may not be preferable - enabled: false few source repo forks: - reason: a package with few repo forks may not be preferable - enabled: false forked source repo: - reason: a forked copy of a popular package may contain malicious code - enabled: true # # type: APIs and permissions # generates new code: - reason: package generates new code at runtime, which could be malicious - enabled: false forks or exits OS processes: - reason: package spawns new operating system processes, which could be malicious - enabled: false accesses obfuscated (hidden) code: - enabled: true accesses environment variables: - enabled: false changes system/environment variables: - enabled: false accesses files and dirs: - enabled: false communicates with external network: - enabled: false reads user input: - enabled: false # # Sandboxing policies # sandbox: rules: # # File system (allow or block accesses to file/dirs) # # ~/ represents home dir # . represents cwd dir # # NOTE: only ONE 'allow' and 'block' lines are allowed # fs: # TODO: customize as per your threat model # block access to home dir and all other locations (except the ones below) block: ~/, / allow: ., ~/.cache, ~/.npm, ~/.local, ~/.ruby, /tmp, /proc, /etc, /var, /bin, /usr/include, /usr/local, /usr/bin, /usr/lib, /usr/share, /lib # # Network (allow or block domains/ports) # # NOTE: only ONE 'allow' and 'block' lines are allowed # network: # TODO: customize as per your threat model # block all external network communication (except the ones below) block: 0.0.0.0 # For NPM packages allow: registry.yarnpkg.com:0, npmjs.org:0, npmjs.com:0