TRIGGER when: user asks 'is everything ready for opening night', user says 'check opening night readiness', session is an opening-night session for a Broadway or West End show. Runs the full 9-point readiness checklist (CLAUDE.md rule 14) and reports PASS/FAIL per check. Takes show_id as argument (e.g. /verify-opening-night giant-2026).
AUTO-LOAD when: user asks about scraping, fetchPage, ScrapingBee, Bright Data, Playwright scraping, adding a new scraper, migrating a fetch call, debugging a 403 or empty response, Cloudflare blocking, paywall cookies, or any script that uses scripts/lib/scraper.js. Contains gotchas, fallback chain, and architecture rules for this project's scraping infrastructure.
MANDATORY for any UI change before push. Sweeps localhost at 5 widths (360-1440), takes element-cropped legibility shots (full pixel resolution — NOT thumbnail), runs structural overflow probe (scrollWidth > clientWidth), and (optionally) runs two-model LLM diff review against reference designs. Writes verdict.json with contentHash. ALWAYS run + Read every element crop at full size + share manifest with user BEFORE pushing UI changes. The pre-push hook blocks deploys until the user gives a plain affirmative (yes/ship it/looks good) in their most recent message — never ask them to copy a hash.
TRIGGER when: user asks 'is everything ready for opening night', user says 'check opening night readiness', session is an opening-night session for a Broadway or West End show. Runs the full 9-point readiness checklist (AGENTS.md rule 14) and reports PASS/FAIL per check. Takes show_id as argument (e.g. /verify-opening-night giant-2026).
TRIGGER when: user has edited scoring logic (review-guards.js, rebuild-all-reviews.js, scoring.ts, engine.ts, data-core.ts), user wants to verify scoring changes, or CLAUDE.md rule 12 section on scoring-logic edits applies. Runs scoring-delta.js to surface T1 flips and total flip count before any push. MANDATORY for all scoring-logic edits — unit tests alone are not sufficient.