Metric Story Reel
Make a short editorial data film from 3–8 dated observations. Preserves the full time window, proportional date spacing, zero-inclusive chart scale, units, source and caveat. Includes actual silent MP4 export and editable scene source.
A little less setup
Make this part of your toolkit.
Give your agent this prompt. You’ll approve access in your browser before it installs a Skill.
Read https://skilly.uk/install/skills/metric-story-reel.md and follow the instructions to install this Skill for my agent.REAL SAMPLE / RELEASED PACKAGE
RELEASED v1.0.0A closer look at the actual output.
Actual normal and boundary MP4 exports: 12 seconds, 1280 × 720, 24 fps, 288 frames, no audio stream.
Read the silent film: text and data
The Reuse Observatory · No. 06 / 2026. Small returns. A growing habit. A fictional neighborhood refill pilot, observed over six months.
Weekly refill activity: Completed refills in the observed week. Counts, not unique customers.
The full picture, over time. The line uses proportional date spacing and a scale that includes zero. Change over the complete observation window: +38 refills / week. Relative change: +82.6%.
A habit takes shape. This sample illustrates a trend. It does not establish why activity changed.
Source: invented demonstration data · The Reuse Observatory
Synthetic demonstration. Equal metric definition across observations · no causal claim · silent film
| Observation date | refills / week |
|---|---|
| 2026-01-05 | 46 |
| 2026-02-02 | 51 |
| 2026-03-02 | 49 |
| 2026-04-06 | 63 |
| 2026-05-04 | 71 |
| 2026-06-01 | 84 |
- Tested runtime
- Node.js 24.19.0 · macOS arm64 · Chromium 153 · FFmpeg 8.1
- Pinned dependency
- Playwright 1.63.0 · local OFL fonts · user-supplied FFmpeg/libx264
- Checks completed
- Input validation, no-overwrite behavior, normal and edge MP4 rendering, ffprobe geometry/frame count, full-file decode and sampled keyframes.
Author normal/edge checks and independent new-brief workflow reviewed; tested scope and limitations apply.
Fixed 12-second landscape 720p films. Deliberately silent. No footage generation, narration or captions. No causal inference or forecasts. Windows/Linux and automatic agent discovery are unverified.
From a good brief to a better result.
Make a short editorial data film from 3–8 dated observations. Preserves the full time window, proportional date spacing, zero-inclusive chart scale, units, source and caveat. Includes actual silent MP4 export and editable scene source. The published package has its own version, tested environments, and included outputs listed alongside this overview. Any earlier draft samples below retain their original verification scope.
What you bring
- Dated numeric observations, metric definition, units and source
- Context, interpretation and caveat; synthetic examples are clearly labeled
What you’ll make
- Silent H.264 MP4 (1280 × 720, 24 fps, 12 seconds)
- Editable deterministic HTML/CSS/SVG scene source and JSON brief
- Eight PNG keyframes and verified render report
The intended workflow
- Prepare the task-specific JSON brief
- Generate deterministic local scene source
- Render and verify the actual MP4 with local Chromium and FFmpeg