agentleFS
tamtree-aitamtree-ai/stick-stage/skills/explainer/SKILL.md

explainer

Skill2 starsChanged 15 days agoMITView on GitHub

Make a narrated 16:9 data-story explainer with StickStage (`format: "explainer"`) — a topic and its data to a checked, voiced MP4 with charts, maps and counters bound to datasets, chapters, and the anti-template variety rules. Use when asked for an explainer, a business or economics "data story", a long-form YouTube video, a video built from a dataset, or "explain X with charts".

22 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use ugc-ads.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • AdviblySkill

    ugc-ads

    Generate complete multi-shot UGC video ads end to end with the Advibly MCP tools. Takes a brand plus an ad angle and produces a realistic AI creator image (gpt-image-2), a 5-shot direct response script, a user-approved storyboard of start frames, and one video clip per shot animated from its frame (Gemini Omni Flash with native spoken dialogue), then assembles the clips into a finished vertical ad. Use whenever the user wants a UGC ad, a creator-style video ad, a product-in-hand testimonial video, or asks to turn a product into a TikTok/Reels-style ad. Also trigger on phrases like "UGC video," "creator ad," "testimonial ad," or "make an ad for my product" when the Advibly MCP is connected.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 3 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 22 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Get the data first: Every number on screen or in a line comes from a datasets table with a source (name, url, retrieved).

    line 15
  2. Distinctive:

    Look for text over captions, crowded labels, and stretches where nothing changes.

    line 33
  3. Distinctive:

    Don't type a number into a figure or a line that isn't in a dataset.

    line 39
  4. Very specific:

    Don't reuse the same scene order, slots and grounds as the last video: the sameness check compares fingerprints when Tamtree passes history.

    line 40
  5. Distinctive:

    Don't put figures where captions go (the bottom ~16 %): use slots.

    line 42

“Common” is measured: kept by at least 6 in 10 of the 22 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Explainers

Guide: docs/explainer.md (format, slots, data binding, focus, chapters, variety rules). Figure params: docs/figures.md (kinds line, bars, counter, map, plus label, callout, progress, image…). Worked example: public/skits/costco-margins/skit.json. Plan: ../planning/2026-10-30-explainer-mode/plan.md.

Steps

  1. Get the data first. Every number on screen or in a line comes from a datasets table with a source (name, url, retrieved). In production Tamtree's data nodes fill them. If you are given no data, use made-up numbers only with "illustrative": true, and say so in the hand-back.
  2. Write the script as scenes. One idea per scene, a chapter title on each (at least three chapters, each 10 s or more), lines under ~20 words, all spoken by the narrator. No cast.
  3. Stage each scene with figures. Pick a slot per figure (full, left/right, center, top, lower-third, inset), bind params to data ("@id.column", "@id.column[key]"), and cue them on words: show, state, set (draw a line on, tick a counter, fly the map with zoom), focus / focus none. Alternate grounds (explainer-paper, explainer-night, and the science themes explainer-dawn / -tide / -cosmos / -lab) and variant (clean / bold) between scenes. Never the retired blueprint grounds (explainer-grid, explainer-drafting*).
  4. Tag spoken numbers. Each number a line says gets "numbers": [{ "text": "13", "data": "@margins.gross_pct[Costco]" }].
  5. Check. pnpm voice:tts <id> (or voice:say), then pnpm check <id>. Fix every error: number-mismatch / number-not-said (the line or the data is wrong: never "fix" data to match a line), variety-kinds (use more kinds), variety-run (switch kind sooner), variety-change (a cue at least every 8 s), variety-hook (open on a figure moving), data-ref, map-zoom.
  6. Render and look as video. pnpm render <id>, then a frame sheet: ffmpeg -i out/<id>.mp4 -vf "select='not(mod(n\,150))',scale=640:-1,tile=3x5" -frames:v 1 out/<id>-frames.png. Look for text over captions, crowded labels, and stretches where nothing changes.
  7. Hand back the MP4, the frame sheet, the chapters from the post text, and which numbers are illustrative. Nothing is approved or posted.

Don't

  • Don't type a number into a figure or a line that isn't in a dataset.
  • Don't reuse the same scene order, slots and grounds as the last video: the sameness check compares fingerprints when Tamtree passes history.
  • Don't put figures where captions go (the bottom ~16 %): use slots.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.