Comment by jonkoops
1 day ago
I really think that the main issue is that NPM itself will execute any script that is in the "postinstall" section of a package, without asking the user for permission. This is a solved problem in other package managers, e.g. PNPM will only run scripts if the user allows them to, and store the allowlist in the package.json file for future reference.
In this scenario, if a dependency were to add a "postinstall" script because it was compromised, it would not execute, and the user can review whether it should, greatly reducing the attack surface.
Wouldn't this just make the number of packages that can be targeted smaller? E.g. I publish a testrunner that needs to install Headless Chrome if not present via postinstall. People trust me and put the package on their allowlist. My account gets compromised and a malicious update is published. People execute malicious code they have never vetted.
I do understand this is still better than npm right now, but it's still broken.
Security is usually full of incremental improvements like that, however. Reducing the scope from all of NPM to the handful of things like test runners would be an enormous benefit for auditors and would encourage consolidation (e.g. most testing frameworks could consolidate on a single headless chrome package), and in the future this could be further improved by things like restricting the scope of those scripts using the operating system sandbox features.
Security is layered, no layer will conclusively keep you safe, but each one make it harder to pierce to the core. For example, the impact of the recent SHA1-Hulud attack would be much less, as compromised packages (that previously did not have any scripts executing at install time), would not suddenly start executing, as they are not allowlisted.
There are a large subset of security problems that are solved by simply eliminating compilation steps typically included in "postinstall". If you want a more secure, more debuggable, more extensible lib, then you should definitely publish it in pure js (rather than, say, Typescript), so that there is no postinstall attack surface.
With type stripping in Node LTS now there's no reason at all to have a postinstall for Typescript code either. There's fewer reasons you can't post a "pure TS" library either.