agentsclimarketplace
Sign in

Youtube video kit

Skill seldonframe/seldonframe/.claude/skills/youtube-video-kit

Produce the complete publishing kit for one keyword-targeted YouTube video (script, thumbnail, title/description/chapters/pinned comment) from its matching SEO page, and wire the videoId back into the page after publish. Invoke with a /best or /tools slug, or a keyword.From its SKILL.md

Install
npx -y skills add seldonframe/seldonframe --skill youtube-video-kit

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

3.2 KB, 753 tokens by cl100k_base, as published. Nobody here has run it

youtube-video-kit — one keyword, one video, one kit

Max records the face; everything else is produced here. The video targets the SAME keyword as an existing page (usually a /best/<slug> page) so the two boost each other: the page embeds the video (videoId seam already live in lib/seo/best-pages.ts + components/seo/lite-youtube.tsx + VideoObject JSON-LD), and the video description links the page.

Inputs

  • A slug (e.g. crm-for-small-business) or keyword. Resolve the matching page + registry data (best-pages.ts contenders / competitor-pricing.ts numbers).

Produce (write everything to docs/strategy/video-kits/<slug>/)

  1. script.md — 6-9 minutes spoken (~950-1400 words), structured:
    • COLD OPEN (first 15s): the pain number, not the topic ("A missed call costs a plumber about $120. Here's the CRM setup that stops that.")
    • The honest framing sentence early: "I build one of these tools, so I'm biased — I'll tell you exactly when the others win." (never-lies ON camera too)
    • One chapter per contender, EXACTLY the facts from the registry (same numbers, same hedges — the video must never contradict the page), each with the honest watchOut.
    • SeldonFrame's chapter states the flat price + the free build-before-signup.
    • CLOSE: "full written breakdown + the calculator is linked below."
    • Short sentences, spoken register, grade-6. Mark [B-ROLL: show <page URL>] cues where screen-recording the live page fits.
  2. thumbnail.png — pull from the live OG endpoint (it IS a thumbnail generator): https://www.seldonframe.com/api/og?kind=best&title=...&aud=...&n=... (or kind=tool for calculator videos). Download it; if a stronger hook exists, note alternate param text. 1200x630 upscales fine for YT.
  3. metadata.md —
    • 3 title options, keyword-FIRST ("Best CRM for Small Business in 2026 — honest ranking (I build one)"), ≤60 chars preferred.
    • Description: keyword in the first sentence, then the page link, the matching calculator link, timestamp chapters, and the standing disclosure line.
    • Chapters (from the script sections, MM:SS placeholders).
    • Pinned comment draft: the one-line honest summary + page link.
    • 10-15 tags.
  4. shorts.md — 3 clip specs (start/end cue by script section, the hook line, vertical caption text) for the repurpose pass. No native shorts content.

After Max publishes

Given the videoId: set videoId (+ videoUploadDate YYYY-MM-DD) on the matching BEST_PAGES entry in lib/seo/best-pages.ts, run the seo specs, commit, push, merge per the house merge method. The page then renders the lite embed + VideoObject schema automatically.

Rules

  • Facts come ONLY from the registries — if the script needs a number the registry lacks, flag it, don't invent it.
  • The disclosure line is non-negotiable and appears in script + description.
  • One video = one keyword = one kit dir. Don't bundle.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most video audio skills give in 753 tokens

Counted across 619 of the 725 authors here whose files we hold, read 2026-09-06

  • Read product marketing context firstin 13 of 619, across 7 files
  • Define the core visual thesis in one sentencein 11 of 619, across 3 files
  • Break the concept into 3 to 6 scenesin 11 of 619, across 3 files
  • Render the smallest working version firstin 11 of 619, across 3 files
  • Start with a low-quality smoke test renderin 11 of 619, across 3 files
  • Add captions for accessibility and engagementin 11 of 619, across 5 files
  • Write the scene outline before writing codein 11 of 619, across 3 files
  • Specify subject, action, camera, style, and moodin 11 of 619, across 5 files
  • Decide what each scene provesin 10 of 619, across 2 files
  • Export one clean thumbnail framein 10 of 619, across 2 files
  • Pick the right tool for the jobin 10 of 619, across 4 files
  • Run the test suite before proposing a fixin 8 of 619, across 7 files

Said here and by no other author read

  • Pull facts only from registries
  • Include the non-negotiable disclosure line
  • Set videoId on the matching page entry
  • Write everything to the kit directory
  • Keep sentences short and spoken

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.