Reading the article, it sounds like this is the other way around? Bitwarden is offering a new API, and OneCLI Agent Vault is integrating with the new API.
Often, we see a feature which is important to free use of a computer as a general-purpose tool locked behind an ever-changing and/or poorly documented API in a closed-source, centralized, de-facto-government-subsidized project.
The power dynamics of that situation are not symmetrical, so it does matter which project(s) are using which API(s) of the other(s).
These tools are useful, but I can't help to feel like they're solving the wrong part of the problem. I really don't have much concern that an agent has access to one of my credentials. Outside of production, most of these credentials are going to be limited in privilege and self-rotatable.
What remains terrifying is the ability to exfil important data or run commands that are malicious.
Lots of assumptions that the article is AI-authored (it could be but I'm not seeing overtly obvious signs - it's quite readable) & a lot of ungrounded assumptions that this is somehow related to Bitwarden integrating AI into their product.
I really thought reading comprehension among HN users was better than this.
There are worse things to mention about OneCLI as it looks like a completely vibe-coded mess, seeing that CLAUDE.md and Claude itself being one of the contributors [0]
Perhaps the most damning discovery is that they don't even do basic dependency pinning [1] [2] which just risks another supply chain attack.
As soon as I saw that, that was everything I needed to know about the project. No security audit whatsoever and Bitwarden believes this is something worth integrating.
I don't understand why this would change any opinion on Bitwarden. Bitwarden offers an API and OneCLI calls the Bitwarden API. How does a 3rd party calling Bitwarden's API say anything about Bitwarden?
Edit: I can see on Bitwarden's site they also call out their support for OneCLI, so I suppose that looks like Bitwarden saying they approve of and recommend OneCLI. But I see recommending an open source solution as a lot less problematic than recommending any other random private startup solution.
OneCLI assumes that the proxy is fully trusted by the agent and it still has authorized access to your accounts.
What happens when the agent environment is breached? All you need is the the fake key + URL of the proxy and that maps to your real keys and you can make authorized requests outside of the agent.
The real keys don't have to be leaked, just the fake ones have to map to the real one; so unless they are rotated, then this is a problem.
Is there anything actually bad with that writing (other than implying that theirs is the first system to solve this)?
AI has been rlhf post-trained to generate text that people find to be clear to read. Are you now looking to reject clear writing just to spite AI labs?
Tangential: Where is Bitwarden on the below roadmap right now? It wasn’t even good to users, but was an alternative to 1Password and others that had long crossed this bridge.
‘Here is how platforms die: first, they are good to their users; then they abuse their users to make things better for their business customers; finally, they abuse those business customers to claw back all the value for themselves. Then, they die. I call this enshittification, and it is a seemingly inevitable consequence arising from the combination of the ease of changing how a platform allocates value, combined with the nature of a "two-sided market", where a platform sits between buyers and sellers, hold each hostage to the other, raking off an ever-larger share of the value that passes between them.’
I don't get what semantic value you're getting by pasting this. It's almost like saying "VC-funded tech = bad", which is an ironic stance to take on this platform.
Is there anything that bitwarden did that is actually bad for you as a customer of theirs?
They switched from a purchase with local vault storage model (where you could sync it to the cloud if you wanted to) to subscription-only with cloud storage they control.
Short of using pass, what are some good alternatives? My main critic of 1Password has been the cost, but it is a very good password manager, and price seems to have gone down... Or at least the dollars has weakened enough that the price has come down for me.
Often, we see a feature which is important to free use of a computer as a general-purpose tool locked behind an ever-changing and/or poorly documented API in a closed-source, centralized, de-facto-government-subsidized project.
The power dynamics of that situation are not symmetrical, so it does matter which project(s) are using which API(s) of the other(s).
What remains terrifying is the ability to exfil important data or run commands that are malicious.
Lots of assumptions that the article is AI-authored (it could be but I'm not seeing overtly obvious signs - it's quite readable) & a lot of ungrounded assumptions that this is somehow related to Bitwarden integrating AI into their product.
I really thought reading comprehension among HN users was better than this.
Perhaps the most damning discovery is that they don't even do basic dependency pinning [1] [2] which just risks another supply chain attack.
As soon as I saw that, that was everything I needed to know about the project. No security audit whatsoever and Bitwarden believes this is something worth integrating.
[0] https://github.com/onecli/onecli/graphs/contributors
[1] https://github.com/onecli/onecli/blob/main/packages/ui/packa...
[2] https://github.com/onecli/onecli/blob/main/packages/db/packa...
I could not be anymore bearish on Bitwarden than before after looking at this and very glad that I don't use them.
Edit: I can see on Bitwarden's site they also call out their support for OneCLI, so I suppose that looks like Bitwarden saying they approve of and recommend OneCLI. But I see recommending an open source solution as a lot less problematic than recommending any other random private startup solution.
What happens when the agent environment is breached? All you need is the the fake key + URL of the proxy and that maps to your real keys and you can make authorized requests outside of the agent.
The real keys don't have to be leaked, just the fake ones have to map to the real one; so unless they are rotated, then this is a problem.
People trying to detect AI and seeing red the moment their AI-sniff test fails are killing discourse.
AI has been rlhf post-trained to generate text that people find to be clear to read. Are you now looking to reject clear writing just to spite AI labs?
So in one part the negative reaction is to staleness. Everything sounds the same.
If it was all the same but dry, terse, and to the point (like technical writing), it wouldn’t be so bad.
But it’s repetitive with an annoying, breathless, get-ready-to-be-impressed voice that many of us find grating.
‘Here is how platforms die: first, they are good to their users; then they abuse their users to make things better for their business customers; finally, they abuse those business customers to claw back all the value for themselves. Then, they die. I call this enshittification, and it is a seemingly inevitable consequence arising from the combination of the ease of changing how a platform allocates value, combined with the nature of a "two-sided market", where a platform sits between buyers and sellers, hold each hostage to the other, raking off an ever-larger share of the value that passes between them.’
- Cory Doctorow
I may be out of the loop, but how was Bitwarden not "good" to users? Does this relate to the recent price increase?
Is there anything that bitwarden did that is actually bad for you as a customer of theirs?