Festivus

Festivus / Changelog

What we shipped, why

A log of every visible change to Festivus. Each entry answers four questions in plain language: what we built, why we built it, what problem it solves for a roboticist, and how it grows the community. No version numbers. No roadmap promises. Just the receipts.

  • Launch
  • Improvement
  • Fix
  1. Launch0xChrisM avatar@0xChrisM

    Compatible-Policy Finder rail on every robot page

    What
    Every robot page now has two new sections. COMPATIBLE is the curated list, what a human vouched for. YOU MIGHT FIND INTERESTING is the algorithm's ranked guess at what might run, with a button to validate the good ones onto the curated list.
    Why
    Festivus search returns 9,296 policies, which is the polite definition of useless if you want to know which five to try tonight. Old search rewards whoever said "Franka" the most in the README, a different question from whether the code will actually run on a Franka. We had the structured fields to answer the real question. We had not gotten around to using them.
    Problem it solves
    Before today, "what code runs on my robot" was a multi-tab afternoon of skimming Hugging Face README files and squinting at action-space specs. The answer was sitting in our database the whole time. We were not telling anyone. People kept asking and we kept shrugging, which was rude.
    How it grows the community
    Every algorithmic card has a small VALIDATE button. Click it and you are dropped into Claude with a pre-written message that promotes the match onto the curated list. One click takes a suggestion from "a machine thought so" to "a human said yes." Discovery feeds validation. Validation feeds curation. The next search is sharper for it.
    Spec 031 →Try it: Franka Research 3 →Try it: Atlas (humanoid) →Try it: Spot (quadruped) →