The author of the article mentioned that they are using it as a honeypot to detect when bots (or rather authors of the bots) implement a work-around for the obfuscation technique. Which is pretty smart!
You could use these shell script versions of pipelines in GHA though, right? There is nothing stopping you from triggering a bash script via a "run" step in YAML.
These reusable actions are nothing but a convenience feature. This discussion isn't much different than any other supply chain, dependency, or packaging system vulnerability such as NPM, etc.
One slight disclaimer here is the ability of someone to run their own updated copy of an action when making a PR. Which could be used to exfil secrets. This one is NOT related to being dependent on unverified actions though.
(re-reading this came across as more harsh than I intended.. my bad on that. But am I missing something or is this the same issue that every open-source user-submitted package repository runs in to?)
What does it mean to put an agent inside of a PG DB? Is PG doing the actual computing/inference (does it tie to GPU), or is PG just storing the state of the agents neural net (is that a thing, it seems like it would be).
Maybe I'm strictly too stupid to even parse the concept, but I don't understand what I'm looking at one bit.
reply