Staged publishing and new install-time controls for npm
github.blogIs any form of code analysis out of the question? Static and dynamic analysis of the code would seem like a promising idea rather than just trying to defer the update and hence the problem.
Seen favorably, staged publishing is a band aid. Seen more realistically I believe that in the long run it will even hurt our efforts for more secure infra.
How could it possibly hurt?
For trusted publishing, it's not a band-aid, it's a significant improvement that kills an entire class of CI takeover publish attacks. I'm sure attackers will find another way but it's a big gap this is closing up.
Nice…maybe will help some of the recent attacks
If maintainers actually use it
This is the biggest question I also had after reading the blog post. Given the recent chain of attacks, wouldn't it make sense to enforce staged publish by default or at least gradually move over to it?
meanwhile pnpm 10.x by default won't donwload packages younger than a day
Is one day enough to find vulnerabilities? Who keeps an eye on new releases? Otherwise the problem continues to exist, just delayed by one day.
There’s almost a dozen cybersecurity companies scanning NPM publishes in real-time and analysing them.
*11.x
Perfect, now we'll start seeing people automate auto publishing because they don't want to explicitly push a button to publish it.