You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
The rejected brand pipeline is still in the repository #18
brand/tools/ is the earlier brand pipeline: generate a handful of plates, then cut every surface out of them. That approach was rejected, and playbook/06-brand/03-pipeline.md records why — a composition briefed for one shape does not survive being cut to another. Surfaces are now drawn at their own native size by forge.
The config drifted along with its disuse. Six of the ten surfaces in surfaces.json name source plates that no longer exist, so python -I surfaces.py fails on them, and it still lists print-a4-landscape, which print-a4-portrait replaced.
The drift is the hazard rather than the dead code. The table reads as authoritative, and the plates it cannot find are the surf-* files that are already approved and already shipping. Anyone reconciling the two concludes the plates are missing and regenerates finished work. That nearly happened on 6 November.
BRAND.md in the same folder already says this folder holds finished assets only and that the generator lives in the private playbook, so the tooling contradicts its own neighbour.
Done when
brand/tools/ is gone — all five files are one unit and reference only each other
brand/plates/wide.png stays. It was listed for removal here because nothing cuts from it; that is the wrong test, since nothing cuts from readme-filter.png either and it is kept deliberately. The owner likes this one.
README.md no longer lists brand/tools/, and describes plates as one per surface rather than one per format
Every surf-* and day-* plate stays, and so do both kept plates, readme-filter.png and wide.png
Activity