← ClaudeAtlas

research-brieflisted

Produce a per-video research brief — competitive scan, viewer-language pull, fact-checking, source list.
Heuresis/YouTube-Agency · ★ 4 · AI & Automation · score 73
Install: claude install-skill Heuresis/YouTube-Agency
# Research Brief ## Purpose Produce the per-video research brief that prepares a script-writer with the audience-language, the competitive landscape, the fact-checked claims, and the source citations needed to write a voice-faithful, Truth-Gate-compliant draft. The brief is the 80% of pre-work behind the 20% of script craft. ## Decision Logic Most scripts fail not because of script-writing skill, but because of insufficient pre-work. The 20/80 inversion: 20% script craft + 80% research / competitive analysis / VOC mining. The brief is the load-bearing 80%. The brief always includes: (a) audience-language pull from VOC for this specific topic, (b) competitive scan (3-5 anonymized competitor archetypes covering this topic — what works, what's missing, where the angle differentiates), (c) fact-checked claim list with sources (everything that needs to survive the 30-second screenshot test), (d) the structural beats hypothesis (hook + sequence + close), (e) the awareness-level + mode declaration. ## Tacit Principles 1. **VOC first, structure second.** Pull verbatim audience phrases on this topic before designing the script structure. The phrases shape the hook + headlines + key beats. 2. **Competitive scan is for differentiation, not imitation.** Read what competitors do, then ask "what's the gap?" The brief surfaces the gap. 3. **Fact-check before script.** Every numeric claim, every "X% of operators…" sourced before drafting. No fabrication. 4. **One angle per brief.** Don't