Comment by curuinor
Comment by curuinor 6 days ago
Comment by curuinor 6 days ago
They delved deep and spent a whole 2 minutes with ChatGPT 4o getting those explanations and apologies in play.
Hey CodeRabbit employees
> The researchers identified that Rubocop, one of our tools, was running outside our secure sandbox environment — a configuration that deviated from our standard security protocols.
This is still ultra-LLM-speak (and no, not just because of the em-dash).
Absolutely. In my experience every AI startup is full of AI maximalists. They use AI for everything they can - in part because they believe in the hype, in part to keep up to date with model capabilities. They would absolutely go so far as to write such an important piece of text using an LLM.
I wonder how many of these intern-type tasks LLMs have taken away. The type of tasks I did as a newbie might have seemed not so relevant to the main responsibilities but they helped me get institutional knowledge and generally get a feel of "how things work" and who/how to talk to make progress. Now the intern will probably do it using LLMs instead to talking to other people. Maybe the results will be better but that interaction is gone.
I think there is an infinite capacity for LLMs to be both beneficial, or negative. I look back at learning and think, man, how amazing would it have been if I could have had a personalized tutor helping guide me and teach me about the concepts I was having trouble with in school. I think about when I was learning to program and didn’t have the words to describe the question I was trying to ask and felt stupid or an inconvenience when trying to ask to more experienced devs.
Then on the flip side, I’m not just worried about an intern using an LLM. I’m worried about the unmonitored LLM performing intern, junior, and ops tasks, and then companies simply using “an LLM did it” as a scapegoat for their extreme cost cutting.
Not a single mention of env vars. Just shifting the blame to rubocop.
They first disabled rubocop to prevent further exploit, then rotated keys. If they awaited deploying the fix that would mean letting compromised keys remain valid for 9 more hours. According to their response all other tools were already sandboxed.
However their response doesn't remediate putting secrets into environment variables in the first place - that is apparently acceptable to them and sets off a red flag for me.
"According to their response all other tools were already sandboxed."
Everything else was fine, just this one tool chosen by the security researcher out of a dozen of tools was not sandboxed.
> putting secrets into environment variables in the first place - that is apparently acceptable to them and sets off a red flag for me
Isn't that standard? The other options I've seen are .env files (amazing dev experience but not as secure), and AWS Secrets Manager and similar competition like Infisical. Even in the latter, you need keys to authenticate with the secrets manager and I believe it's recommended to store those as env vars.
Edit: Formatting
You can use native authentication methods with Infisical that don't require you to use keys to authenticate with your secrets manager: - https://infisical.com/docs/documentation/platform/identities... - https://infisical.com/docs/documentation/platform/identities...
They weren't published together. They managed to get the researchers to add CodeRabbit's talking points in after the fact, check out the blue text on the right hand side.
https://web.archive.org/web/diff/20250819165333/202508192240...
The LLM tics are strong in this writeup:
"No manual overrides, no exceptions."
"Our VDP isn't just a bug bounty—it's a security partnership"