Forecut
AI trend forecasting for garments, footwear and detailing.
Live site: https://forecut.in/
index.html: homepageaw26.html, ss27.html, aw27.html, ss28.html, aw28.html: season reportsbuild/: data and scripts that generate the season pages (see build/README.md). Rebuild with cd build && STATIC=.. python3 gen_seasons.py.
Updated monthly.
Reports and Journal (added Oct 2026)
build/journal_data.py → journal.py: Journal articles (journal.html + one page per article).build/reports_data.py → reports.py: Core Item Updates, Design Capsules, India Colour Forecast, Textile & Craft Sourcing Guide, reports.html hub. Flat sketches are pre-rendered with node + flats.js.- Statuses always come from
india_data.py, so reports update when the India forecast is refreshed. Run sh build.sh from build/.
- Never copy text, palettes, data or images from paid forecasting services (e.g. WGSN) into these pages; use them only as format inspiration.
Launch locks
build/site.json → "locked" lists seasons that stay locked (tabs, lifecycle chips, tracker columns, India calendar/switcher, AW27/SS28/AW28 detail sections). At official launch set it to [] and run sh build.sh.
Golden rule: never delete
Details (global and India), Journal articles, Brand Watch entries, photos and season pages are only ever added or updated, never removed.
- Fading items stay, marked "Fading"; past seasons are archived (rollover.py keeps the page and saves past statuses), not deleted.
python3 keep_check.py (run by build.sh) fails the build if anything that existed in the last commit is missing. If it fails, restore the item; do not bypass the check.
Flat sketch images
Upload flat-<slug>.png (or flat-<slug>-<season>.png) to images/. They are padded to 1000×1000 on white (never cropped), replace the code-drawn sketch on cards without a photo, and appear as a "Flat sketch" toggle next to photos in the detail viewer. Prompts: season-prompts/ss27-flat-sketch-prompts.csv.
Independent India season reports
build/india_season_reports.json owns the five curated season edits, with 20 unique development details each, pricing, proposed merchandising offers, documented collection stories, innovation validation and sources. The previous full detail catalogue is preserved in india-reference-library.html and the original Complete-Report.pdf files. Do not delete either edition.
In build/, run STATIC=.. python3 gen_india.py to produce the season hub, five dedicated pages and reference archive. Run python3 india_season_pdfs.py after changes to report content or images, then regenerate the pages to refresh PDF page counts. Run seo.py normally, or set ONLY to the changed filenames and SITEMAP=1 for focused metadata and sitemap updates. Run both test_india.py and test_india_season_pages.py using Python 3.12 with ReportLab, Pillow and pypdf; Node must be on PATH for script checks. PDF fonts use the same FORECUT_PDF_FONTS override as the extended PDF builder.
Generated concept prompts and asset provenance are recorded in build/india_detail_image_provenance.json. Reference images illustrate design vocabulary; generated images are original concepts. Neither is future runway evidence. 2028 content is conditional. Retailer snapshots are dated and separate from Forecut's proposed future price bands and offers.
