lp-brief
Landing Page Brief — Orchestrator
Communication — Step between strategy and design. Coordinates audit anchoring, hypothesis generation, architecture, per-section specification, asset slotting, and hand-off prompt composition into a single approved brief.
Core Question: "Could a designer (or Claude Design) build the right page from this brief without a single follow-up question?"
Critical Gates — Read First
- Do NOT generate a brief without brand artifacts. Missing
brand/BRAND.mdorbrand/DESIGN.md→ returnNEEDS_CONTEXT. Brief depends on tokens, voice rules, sacred elements. - Do NOT skip the conversion rubric. Every section spec is gated by lp-optimization's craft rules (4-U headline, message match, CTA psychology, social proof placement, objection handling, form-field discipline). Brand-good but conversion-bad = failure.
- Do NOT propose changing sacred elements. Logo geometry, primary palette anchor, tagline wording, signature treatments are "do not touch" rails, not options.
- Do NOT exceed the brief length envelope. Useful brief is 250–500 lines. <250 = insufficient depth (designer asks follow-ups). >500 = bloat (designer skims, misses spec). Critic enforces.
- Do NOT inline the full skill chain. If project has a shared chain doc (e.g.,
growth/page-redesigns/_prompts.md), reference by section header; add page-specific overrides only. - Do NOT inject placeholder testimonials, fake logos, or pretend numbers. If a proof asset isn't real, spec it ("Customer logo grid, 6 cells × 60px") and note "delete cell if not real" — never fabricate.
Philosophy
Operates between strategy and design. By invocation, WHY (hypothesis) and WHO (audience) should be stable; the brief turns those into HOW (architecture + section spec + asset slots) precise enough for execution. Conversion rubric from lp-optimization is internalized — every section evaluated against the same craft rules used to audit existing pages, but applied at brief time so the page is built right rather than audited later.
More from hungv47/marketing-skills
humanize
Strips AI patterns, injects brand voice, and compresses existing text so it reads human-written. Targets 15%+ word reduction with zero idea loss. Produces `.agents/skill-artifacts/mkt/content/[slug].humanized.md`. Not for writing new copy (use copywriting). For brand voice reference, see brand-system. For SEO compliance, see seo.
20lp-optimization
Audits a landing page for conversion — analyzes hero, CTA, social proof, objection handling, and page flow. Produces specific copy and structure change recommendations. Not for designing A/B test variants or full site SEO audits (use seo). For brand identity and design tokens, see brand-system. For content strategy, see campaign-plan.
20brand-system
Builds brand identity systems as three artifacts — BRAND.md (story, voice, positioning, archetype), DESIGN.md (AI-readable design system with palettes, tokens, components, motion), and ASSETS.md (per-platform production inventory with auto-scanned checkboxes for what's done vs. still needed). Not for writing marketing copy (use copywriting) or mapping user flows (use user-flow). For campaign planning, see campaign-plan. For audience research, see icp-research.
20seo
Audits and plans search visibility — keyword research, on-page optimization, technical SEO, link building strategy, and AI search optimization. Produces `.agents/skill-artifacts/mkt/seo-[mode].md`. Not for landing page conversion (use lp-optimization) or writing copy (use copywriting).
19vn-tone
Polishes already-translated Vietnamese text so it reads natively in a target register (báo chí, semi-casual, bro, or pop-marketing). Fixes pronoun drift, missing particles, literal idioms, passive-voice calques, typography, and corporate translationese. Post-translation polish pass only — does NOT translate from other languages. For tone work on English/other languages, see humanize. Produces `.agents/skill-artifacts/mkt/content/[slug].vn-tone.md`.
18copywriting
Writes and evaluates persuasive copy — headlines, hooks, CTAs, taglines, and full-page section copy with rubric scoring, annotations, and ranked alternatives. Produces inline annotations or `.agents/skill-artifacts/mkt/content/[slug].copy.md`. Not for editing AI-sounding text (use humanize). For brand voice guidelines, see brand-system. For landing page conversion audits, see lp-optimization.
18