Answering it for one change means reading five sources that never reference each other — and on two of the repo's live changes those sources currently disagree, so the answer isn't just scattered, it's wrong.
Despite all our tooling, most of us still ask "what's going on?" and "what are we working on?" and that's surprisingly hard to answer. Start with a change as the unit of currency.
A read-only page at /changes/[id] that assembles one OpenSpec change from the repo and git: intent, a stage rail, every outcome with the kind of evidence behind it, a dated history that keeps its silences, and a band comparing what the change claims against what the record shows. Nothing is written back and no field has to be remembered.
PI 4 — solves a regular problem, would use frequently. The question this answers is one Katy asks every time a change is picked back up after a gap, which in this repo is often: tell-the-story sat untouched for 31 days and then produced four pull requests in a single day. Coming back to it meant reconstructing where it stood from five sources by hand. Not a 5, because it is not a daily tool. It is a re-entry tool — read hard at the moment you return to something, ignored while you are in flow on it. The honest frequency is per-change, not per-day. The strongest personal signal is that it caught a mistake in its own design while being built: the evidence split for tell-the-story had been read by hand as 5 of 8 outcomes test-backed, and derived from the record it is 8 of 8. That error had already reached the proposal, the design and the Figma frames.
ARTIFACTS
On the hub
- Change page — where it stands, and how it got there
