I think im missing something, don't all harnesses (opencode, pi, etc) already do stuff like "retry"? As far as I can see, when a tool call fails in either, the model gets the error back to correct.
Nice find. Its a bit strange that the PRs listed there, are not present at all in the coinbase repo. Seems like the attack was directed there, but I also did not hear anything from Coinbase on this.
eg. Target their NPM and PYPI tokens, so they can push compromised packages.
Docker Desktop includes the easy to run Docker Engine / Docker Machine. I think is fair to assume that most of the revenue is not from users that want a GUI but from users that want a stable Docker Engine experience.
GitOps (as in ArgoCD style workflows) is not nice. Deploying from code, and storing all infra, etc as code is great, but this pull based approach creates IMHO a really decoupled and inconvenient deployment pipeline.
EDIT: OK, I see the main trick is to use PATH instead of shims.
> rtx does not use shims and instead updates PATH so that it doesn't have any overhead when simply calling binaries
There is a good reason `asdf-vm` uses shims, and is that it does not have to interplay or worry about other tools that set PATH and tools that need a reference to an executable could be simply set to `~/.asdf/shims`.
Ill take it for a spin, but this choice might have a lot of consequences that are not easy to foresee. A good example is `direnv` which as you mention in the README now requires to be set in `.envrc` and then disable global `rtx` hook I guess.
I only see 1 user confirming it and he had a special or odd disk setup (merged disks which also do not show as merged, might be the source of the problem).