Launch HN: Patched (YC S24) – AI workflows for post-code tasks

  • Certainly agree about the bottleneck around DevSecOps. I wonder if this might be useful for the typical "scan everything, patch everything, SBOM for everything" loop that typically lands on DevOps teams and becomes a classic devops tug-of-war situation.

    I'll be trying the free-tier to accomplish this on a hobby project at some point soon, will try to provide feedback! Proof of patching, SBOM, compliance stuff is IMO one of the best moats existing companies have against newcomers - developers typically _hate_ security patching work - so there's money in that use-case for sure!

    Oh, edit: Congrats on the launch!

  • Congrats on the launch; the idea and process is a good one, but the results so far are less impressive.

    For a simple PoC typescript CLI tool I had, AutoFix decided to do this several times:

    https://i.postimg.cc/1z4Fs663/Screenshot-2024-10-31-at-21-37...

    In a straight forwards repo with no README it created a reasonable starting point.

    Yet in a more complex repo with an existing and comprehensive README, it decides to replace it completely with a simple one, removing key insights. I suspect the existing README isn't considered at all, making this kind of patch incompatible with most workflows (i.e. creating a README once isn't particularly onerous, keeping it up to date is).

    This may be a project to watch, but I'm disinclined to use it at the moment.

  • I thought "post-code" was referring to an era when we don't code anymore, because AI does it all. Made my heart jump a bit.

  • Congrats on the launch! One piece of feedback - I find the tagline confusing. I had no idea what "post-code tasks" meant until I clicked around and saw a few examples.

  • It's cool that there's a chat interface in addition to the PR's. I find that for most reviews a little bit of back and forth is helpful.

  • Its seems very nice, I will try it

  • still not sure what the value add is here

  • “AI workflows”

    read: we sprinkled some LLM prompts and some job orchestration