forge-mirror-scaffoldlisted

# /forge-mirror-scaffold — generate a Forge mirror page from a URL Walks a live URL via `crawler --capture-reference`, extracts the SSR-visible content, and emits a valid `cms/<slug>.json` that passes `forge build` (0 strict findings). Designed for the PRIORITY 3 rotation (prosperityclub.com / plausiden.com / sacred.vote / Stripe / Linear / Vercel / GitHub / Notion / Anthropic / Render / Fly) where each page needs a working approximation rapidly. ## Why this exists This-session experience: 11 Forge-static mirror pages hand-authored at ~150 LOC each. Each one followed the same pattern (`image_hero` with eyebrow + verbatim title + verbatim lede + CTA → `paragraph` disclaimer → `pull_quote` from a verbatim live quote → `kv_pair` 4-6 items mapping the live features → `heading` + `paragraph` → final `call_to_action`). Each one tripped the same `label_consistency` gate that needed site-prefixed hrefs. That's ~1,600 LOC of mechanical CMS authoring that compresses to one command per site if a skill walks the SSR + emits the scaffold. ## Steps ### 1. Capture the live site ```sh sudo -u paul /home/paul/projects/PlausiDen-Crawler/target/debug/crawler \ --capture-reference "$URL" \ --out-dir /tmp/captures ``` The runner emits `<slug>.png`, `<slug>.html`, `<slug>.styles.json`, and `manifest.json` in `/tmp/captures/<slug>/`. Confirms multi-viewport capture works against the target. ### 2. Extract content from the captured HTML Read `/tmp/captures/<slug>/1280.html` and extract