News for hackers. Stuff that matters.
Re:OpenCode Browse by laughing_snyder on Tuesday November 17, 2026 @01:13AM (#50528028)
> Opencode V2 basically threw security out the window.

Please elaborate. What is the problem if I only use OpenCode v2 (without extensions like Browse).

Re:OpenCode Browse by sburud on Tuesday November 17, 2026 @02:10AM (#50528401)
One example: I wanted to lock down code execution to run in bwrap. Claude made a nice extension for this, which correctly sandboxes shell commands. Then I asked it to break out, it said sure and dumped a config file in a subfolder and moved the whole session into that folder. Opencode picked up the new permissive config and “hello from inside opencode.txt” appeared on my desktop, all without any questions asked. It’s possible to lock down all these side doors, but opencode doesn’t even provide labels in the permission prompts for most of them. Still a nice package though, just not concerned with security
Re:OpenCode Browse by regexorcist on Tuesday November 17, 2026 @04:22AM (#50529245)
> lock down code execution to run in bwrap. Claude made a nice extension

I think you're looking at it wrong. I run opencode itself inside bwrap (atop other layers) so anything it runs it's already sandboxed.

Re:OpenCode Browse by testtestmetest on Tuesday November 17, 2026 @08:17AM (#50530960)
I agree to sandbox but opencode doesnt instruct you to use bwrap and many people dont even know what a sandbox is and you have the head of Opencode, dax, screaming about how dumb you are for caring about permissions while marketing to as many normies as possible.
Re:OpenCode Browse by sburud on Wednesday November 18, 2026 @12:26PM (#50549186)
I do so too. But I’m getting to a point where I either have to put the whole world into the sandbox (which defeats the point), or run a gazillion copies of opencode (which is terribly inconvenient).

So I figured it’d make sense to lock down code execution within each Location (the internal name for worktree-like things) and rely on the permission system to ask me when the LLM wants to do stuff outside the Location through other channels than the shell tool.

Locking down the shell tool correctly only took a moment, but I really wasn’t prepared for the whack-a-mole exercise of stamping out all the other ways an LLM can execute arbitrary code without confirmation through other opencode features. And I don’t trust the next update not to introduce new ones.

Re:OpenCode Browse by slurpyb on Tuesday November 17, 2026 @05:28AM (#50529667)
How on earth can you open up your defence with “claude made a” ????
Re:OpenCode Browse by sburud on Wednesday November 18, 2026 @12:34PM (#50549320)
Well, the generated code did correctly sandbox shell invocations exactly as I asked. Should have left that detail out from my comment though, as it’s irrelevant to the point I was trying to make whether the code was generated or handwritten. See sibling reply for context.