`build-queue/done/30-research-androoagi.md`

18 lines
# Task 30 — RESEARCH: @androoagi design north-star (videos + posts)

Read SPEC.md + ~/.claude/skills/abramsego-design-research/SKILL.md. Steve directive: "look into
every site and every post" of the design north-star. Budget-aware: cap this task at ~15 min; it's
best-effort research, not perfection.

1. Ensure yt-dlp: `command -v yt-dlp || brew install yt-dlp` ($0, local).
2. Try pulling @androoagi's TikTok profile videos (newest ~5-8) to ~/Videos/design-research/androoagi/
   via yt-dlp. Also try their X profile (x.com/androoagi) posts via WebFetch. If TikTok blocks local
   IP: note it in the memo and move on — do NOT burn the budget fighting anti-bot (Browserbase
   fallback is task-31's owner if needed, ≤$0.50 total, log to cost-tracker).
3. For any videos obtained: `ffmpeg -vf fps=1/3` frames → read 6-10 frames → describe panel layout,
   palette, density, motion patterns of the "AI Agent Environment Tour" mission-control style.
4. Write findings to memos/design-research.md (create; section "androoagi"): what it shows, 3-7
   concrete patterns worth copying mapped to AbramsEgo panels, what NOT to copy, an "Apply next" list.
   State every cost line ($0 (local) or the metered charge).
5. Commit `research: androoagi design north-star notes`. Local only, read-only web, no posting.