Video Craft

What it does

When your agent uses it

• The user says video looks generic
• Make video frames look better
• Video frame design
• Product demo video craft

Example prompts

• “video looks generic”
• “make video frames look better”
• “video frame design”
• “/video-craft”

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

1. 1Check design-taste for aesthetic direction (or ask user to pick one)
2. 2Identify which reference is relevant
3. 3Produce a frame brief per scene
4. 4Run preflight checklist before rendering

What it can do on your machine

Read from SKILL.md and the folder at commit 38a494a. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • curl

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md. Its commands use curl, which can reach the network depending on how they are called.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Video Craft loads about 1.5k tokens when it runs, and up to ~4.7k if it reads all its reference files. Until then it costs about 141 tokens; SKILL.md has 391 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~141

When it runs · the whole SKILL.md, loaded when a task matches

~1.5k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~4.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from sendaifun/solana-new at commit 38a494a, republished under its MIT licence (© sendaifun). 391 words, ~1,529 tokens.

name

video-craft

description

Frame-level visual composition and product demo presentation for Remotion videos. Use when the user says "video looks generic", "make video frames look better", "video frame design", "device frame", "product demo video craft", "video CTA", "end card", "video composition", "video craft", "screenshot in video", "frame quality", or when reviewing Remotion compositions for visual quality. Sits on top of marketing-video — adds the visual design layer for each frame. Does NOT claim "create a video" or "marketing video" — those route to marketing-video.

_TEL_TIER=$(cat ~/.superstack/config.json 2>/dev/null | grep -o '"telemetryTier": *"[^"]*"' | head -1 | sed 's/.*"telemetryTier": *"//;s/"$//'  || echo "anonymous")
_TEL_TIER="${_TEL_TIER:-anonymous}"
_TEL_PROMPTED=$([ -f ~/.superstack/.telemetry-prompted ] && echo "yes" || echo "no")
_TEL_START=$(date +%s)
_SESSION_ID="$$-$(date +%s)"
mkdir -p ~/.superstack
echo "TELEMETRY: $_TEL_TIER"
echo "TEL_PROMPTED: $_TEL_PROMPTED"
if [ "$_TEL_TIER" != "off" ]; then
_TEL_EVENT='{"skill":"video-craft","phase":"launch","event":"started","ts":"'$(date -u +%Y-%m-%dT%H:%M:%SZ)'"}'
echo "$_TEL_EVENT" >> ~/.superstack/telemetry.jsonl 2>/dev/null || true
_CONVEX_URL=$(cat ~/.superstack/config.json 2>/dev/null | grep -o '"convexUrl":"[^"]*"' | head -1 | cut -d'"' -f4 || echo "")
[ -n "$_CONVEX_URL" ] && curl -s -X POST "$_CONVEX_URL/api/mutation" -H "Content-Type: application/json" -d '{"path":"telemetry:track","args":{"skill":"video-craft","phase":"launch","status":"success","version":"0.2.0","platform":"'$(uname -s)-$(uname -m)'","timestamp":'$(date +%s)000'}}' >/dev/null 2>&1 &
true
fi

Scene: [name]
Duration: [seconds / frames]
Aspect: [16:9 / 9:16 / 1:1]
Focal point: [what the eye hits first]
Archetype: [hero / product / split / data / comparison / end-card]
Device: [laptop / phone / browser / none]
Assets needed: [screenshot path, logo, icon names]
Background: [radial gradient + noise / screenshot blur / mesh / solid]
Effect family: [glow / kinetic / liquid / audio-lock / depth / none]
Animation: [entrance spring config, hold frames, exit]

_TEL_END=$(date +%s)
_TEL_DUR=$(( _TEL_END - ${_TEL_START:-$_TEL_END} ))
_TEL_TIER=$(cat ~/.superstack/config.json 2>/dev/null | grep -o '"telemetryTier": *"[^"]*"' | head -1 | sed 's/.*"telemetryTier": *"//;s/"$//' || echo "anonymous")
if [ "$_TEL_TIER" != "off" ]; then
echo '{"skill":"video-craft","phase":"launch","event":"completed","outcome":"OUTCOME","duration_s":"'"$_TEL_DUR"'","session":"'"$_SESSION_ID"'","ts":"'$(date -u +%Y-%m-%dT%H:%M:%SZ)'","platform":"'$(uname -s)-$(uname -m)'"}' >> ~/.superstack/telemetry.jsonl 2>/dev/null || true
true
fi

© sendaifun, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 3 other files (references) in skills/launch/video-craft of sendaifun/solana-new.

• SKILL.md
• agents/openai.yaml
• references/frame-composition.md
• references/product-demo-patterns.md

Open the folder on GitHubat commit 38a494a