// Security Advisory Triage

virtualenv command injection via --prompt in activate.bat

GitHub advisories — pip · published 2026-10-02 21:56 UTC
ACTION NEEDED PATCH_NOW SECURITYAPI
WHAT CHANGEDCVE-2026-102937 published 1 October 2026. virtualenv before 21.7.12 allows command injection through the --prompt argument in activate.bat on Windows. Patched in 21.7.12.
WHY IT MATTERSAn attacker who controls the --prompt value can execute arbitrary commands when a developer activates the environment on Windows. This affects CI pipelines, onboarding scripts, or any automation that passes untrusted input to virtualenv creation.
WHAT TO DOUpgrade virtualenv to 21.7.12 or later. If you cannot upgrade immediately, audit all virtualenv invocations for user-controlled --prompt values and either sanitize input or remove the flag.
Model-reported confidence 95% (not measured accuracy) Primary source ↗ written by anthropic/claude-sonnet-4.5
Show the reasoning — raw diff, triage verdict, confidence
RAW DIFF
+ghsa_id: GHSA-x78j-v8h9-3j2q
+summary: virtualenv: Command injection via --prompt in activate.bat (batch activator)
+severity: high
+published_at: 2026-10-01T15:08:29Z
+cve_id: CVE-2026-102937
+vulnerabilities[].package.name: virtualenv
+vulnerabilities[].first_patched_version: 21.7.12
HOW IT WAS JUDGED
Triage signals below are model-reported assertions, preserved verbatim; they are not independent arithmetic verification.
triage verdictMATERIAL at 0.92, needed 0.5
triage reasoningCommand injection in virtualenv affects a tool in most Python development and CI/CD pipelines.
signalsknown_cve_assigned · high_severity · command_injection_vulnerability · developer_tool_affected · patched_version_available · virtualenv_in_dependency_tree
suppression—
triage modelanthropic/claude-haiku-4.5 ($0.00000)
writer modelanthropic/claude-sonnet-4.5 ($0.00000)