[GREEN] Engine researches first, then designs, then builds.
This keeps ChatGPT as messenger, not designer.
/var/lib/voidcomms/engine-missions/20260509-071855
VOIDCOMMS ENGINE BUILD MISSION Research VoidComms first before designing anything. Read: - current live pages - public reports - existing rules - existing prompts - learning reports - previous rebuild mistakes Then decide your own version of the site. Do NOT copy old layouts. Do NOT use generic SaaS design. Do NOT make placeholder pages. Do NOT let ChatGPT taste decide the design. You decide: - colours - spacing - typography - wording - hierarchy - motion - layout - imagery direction - page structure - mobile flow - public/private UX VoidComms should feel like: - a public intelligence system - a future internet control layer - a commercial infrastructure company - a proof and decision engine - a serious operator platform The site must clearly explain: - public reports - private owner decisions - protected actions - website factory - prediction pressure - worldwide localisation - learning analytics - Engine decides, owner approves Before building: 1. Research current site 2. Produce design brief 3. Produce visual style report 4. Produce page-by-page plan 5. Produce risk review 6. Explain top design choice and why 7. Reject weak directions 8. Only then generate the site Required pages: - homepage - public reports - public decisions - proof - website factory - predictions - worldwide - phone hub - contact - learning - design report Rules: - each page must have unique purpose - no repeated templates - no fake metrics - no fake users - no fake guarantees - no hidden private controls - public can read - owner decides Final output required: - design_manifest.json - page_plan.json - risk_review.json - full build package - deployment recommendation - why this version is better than the old site
Do not overwrite the live site until the Engine build package is reviewed.