Rapid Broadening Of Cooldown/Age-Gating Support Across Package Managers

Issue 83 Edition 2026-03-24 5 min read

General

Sources: 1 • Confidence: High • Updated: 2026-04-13 03:53

Key takeaways

An Andrew Nesbitt article published March 4 reviews the current state of dependency cooldown mechanisms across packaging tools.
Relative duration support for pip’s --uploaded-prior-to has been requested but is not implemented.
A supply chain attack affecting LiteLLM prompted renewed focus on dependency cooldowns.
Between September 2025 and February 2026, pnpm, Yarn, Bun, Deno, uv, pip, and npm added or highlighted minimum dependency age controls.
Dependency cooldowns reduce risk by delaying installation of newly updated dependencies for a few days to allow time for subversion to be detected.

An Andrew Nesbitt article published March 4 reviews the current state of dependency cooldown mechanisms across packaging tools.
Between September 2025 and February 2026, pnpm, Yarn, Bun, Deno, uv, pip, and npm added or highlighted minimum dependency age controls.
Dependency cooldown support is described as surprisingly well supported, with a recent flurry of activity across major packaging tools.

Relative duration support for pip’s --uploaded-prior-to has been requested but is not implemented.
pnpm and Yarn support dependency age gating and include exemption mechanisms for trusted or preapproved packages.
A workaround for pip’s lack of relative dates is to use a scheduled cron job to keep an absolute date in pip.conf up to date.

A supply chain attack affecting LiteLLM prompted renewed focus on dependency cooldowns.
Dependency cooldowns reduce risk by delaying installation of newly updated dependencies for a few days to allow time for subversion to be detected.

Relative duration support for pip’s --uploaded-prior-to has been requested but is not implemented.

What specific indicators of compromise, blast radius, and downstream impacts were observed in the LiteLLM supply chain attack?
What are the exact semantics, defaults, and enforcement points of the minimum dependency age controls across pnpm, Yarn, Bun, Deno, uv, pip, and npm?
How widely are dependency cooldown controls actually deployed in real CI/CD pipelines (versus being available but rarely enabled)?
What cooldown duration is used in practice, and how do teams balance cooldown windows against patch SLAs for legitimate security fixes?
How are exemption mechanisms governed (criteria, auditing, review cadence), and what failure modes occur if exemptions are overused?

Rising default use of dependency age gating could shift security spend toward supply chain tooling, CI policy management, and auditing, as teams operationalize cooldowns and exemptions across multiple language ecosystems.
Package managers adding or emphasizing minimum dependency age controls suggests ecosystem level convergence on supply chain risk mitigation, potentially increasing demand for governance features that standardize policies across heterogeneous build systems.

Public defaults or guidance from major package managers or platform providers that make minimum dependency age controls recommended or enabled by default, plus documentation on exemption governance and auditability.
Evidence of broad enterprise adoption such as CI templates, reference pipelines, or platform features that enable cooldowns centrally, and metrics showing reduced exposure to newly compromised dependencies.
Implementation of pip relative duration support or widely used tooling that closes the usability gap, indicating sustained user demand and operational integration rather than niche experimentation.

Surveys or telemetry indicating cooldown controls exist but are rarely enabled in real CI/CD, with teams citing unacceptable delivery friction or patch SLA conflicts.
High profile incidents where attackers evade cooldowns or exemptions are routinely misused, undermining the mitigation value and reducing motivation to deploy age gating.
Package managers roll back, deprecate, or stop highlighting minimum dependency age controls, or fail to align semantics, leading to fragmented behavior and limited cross ecosystem standardization.