youtube-branding-738278 — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited youtube-branding-738278 (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
The channel-identity lego-block of the marketing-guru family. Before you ideate and deploy videos (marketing-ideate → marketing-deploy), a channel needs a face: this skill turns a name + slogan + niche into a full, upload-ready brand kit.
| Asset | Size | YouTube use |
|---|---|---|
banner.png | 2560×1440 | Channel banner (art + exact wordmark in the safe area) |
profile.png | 1024×1024 | Channel profile picture / avatar |
logo.png | 1024² transparent | Branding master (transparent PNG) |
logo_512.png | 512² transparent | Watermark to overlay ON videos |
brand.md | — | Channel keywords + description (copy-paste) |
Where it sits in the growth loop (see marketing-guru):
[ youtube-branding ] → IDEATE → BACKLOG → DEPLOY → MEASURE → LEARN
(channel setup, (what to make & how it does, per channel)
once / on rebrand)The mechanical pipeline lives in the studio package (studio/marketing/brand.py) and is driven by the `studio brand <spec.json>` CLI command. It:
in YouTube's safe area (never trust the image model to spell),
You (the agent) author the brand identity — palette, emblem concept, banner scene — into a brand-spec JSON. That creative judgement is the whole job; the CLI is just I/O.
cd /Users/dasein/dev/slope-studio
source .venv/bin/activate 2>/dev/null || { uv venv && source .venv/bin/activate && uv pip install -e ".[fal]"; }
grep -q FAL_KEY .env && echo "FAL_KEY ok" || echo "MISSING FAL_KEY — set it in .env (needed for real art)"You need: channel name, slogan, and the niche/aesthetic. If the aesthetic is unstated, infer one from the niche (don't over-ask) — pick a palette + a single repeatable motif/emblem that ties avatar↔logo.
Copy example-spec.json and rewrite every field. Rules that make the output good:
no text, no letters, no words — the wordmark is added later by Pillow, so the art must be clean.(rembg cuts that background out). Keep it legible at watermark size — few elements.
centre** (push detail to the left/right edges) so the wordmark has room.
primary (deep bg), accent (title glow + rule), highlight(slogan), text (title fill), each [r,g,b].
are the channel copy. `slug` is the output folder under runs/_brand/.
# validates the spec + full Pillow/rembg pipeline with offline placeholder art, $0.
# ⚠ writes to runs/_brand/<slug>/ and OVERWRITES — set a THROWAWAY "slug" for the test.
studio brand myspec.json --provider stubstudio brand myspec.jsonThree Nano Banana stills @ $0.039 = ~$0.117. Output lands in runs/_brand/<slug>/.
Always view banner.png, profile.png, logo.png — verify:
logo.png corner alpha = 0);Re-roll a weak asset by tweaking its prompt in the spec and re-running (a full run re-rolls all three stills; accept the small re-spend, or dry-run with stub first).
Report the asset paths, the keywords, and the description, and where each goes on YouTube (banner / picture / Settings→Basic info). Offer the video watermark overlay:
# overlay logo_512.png bottom-right at ~10% width on a finished video
ffmpeg -i in.mp4 -i runs/_brand/<slug>/logo_512.png -filter_complex \
"[1]scale=iw*0.10:-1[wm];[0][wm]overlay=W-w-40:H-h-40" -c:a copy out.mp4name/sloganare always Pillow-overlaid; that's why style must forbid text.
see film-maker studio thumbnail) needs a verified YouTube channel.
from a plain background, so author logo_prompt with a clean subject on a flat bg.
with graceful fallback.
film-maker; decidewhat to post and run the loop with marketing-guru.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.