agentleFS
45ck45ck/video-evaluator/skills/storyboard-transitions/SKILL.md

storyboard-transitions

Skill2 starsChanged 7 months agoMITView on GitHub

Infer frame-to-frame transitions from storyboard image diffs and OCR deltas, then write an evidence-backed transition manifest.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    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:

    Use this after storyboard-ocr when an agent needs sequence-level change evidence instead of isolated sampled frames.

    line 8
  2. Very specific:

    Run after storyboard-ocr. It benefits from hybrid extraction because same-screen sampling metadata can support state-change interpretation.

    line 42
  3. Distinctive:

    Do not infer user intent or causal actions from transitions alone.

    line 57

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “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

Storyboard Transitions

Use this after storyboard-ocr when an agent needs sequence-level change evidence instead of isolated sampled frames.

This skill owns:

  • image-level diff between adjacent storyboard frames
  • OCR text additions and removals
  • coarse transition kinds such as screen-change or state-change
  • overlap/shared-line signals for same-screen heuristics
  • heuristic transition labels with confidence
  • storyboard.transitions.json

Inputs

Provide either storyboardDir or ocrPath:

{
  "storyboardDir": "./output/storyboard",
  "threshold": 0.02
}

threshold is the normalized visual-diff sensitivity and defaults to 0.02.

Outputs

The skill writes storyboard.transitions.json with adjacent-frame transitions. Each transition includes frame indices, timestamps, visual diff percentage, OCR overlap, shared line count, added/removed lines, transition kind, confidence, and evidence notes.

Sequencing Guidance

Run after storyboard-ocr. It benefits from hybrid extraction because same-screen sampling metadata can support state-change interpretation. Run storyboard-understand after this when you want a higher-level flow summary that can incorporate transition evidence.

Interpretation Notes

screen-change suggests a broader view change. state-change suggests the same shell with changed content. scroll-change and dialog-change depend on stable anchors, OCR regions, and visual deltas. uncertain is expected when OCR was rejected or adjacent frames have too little shared evidence. Confidence is heuristic, not a probability.

Abstention Rules

Do not infer user intent or causal actions from transitions alone. Abstain from precise workflow claims when transition confidence is low, OCR quality is rejected, or frame sampling skips the moments between two states.

Failure Modes

The runner needs existing OCR output and accessible frame images. Missing OCR, missing images, or ffmpeg failures during diff preprocessing will fail the run. Very small threshold values can over-label visual noise; high values can hide subtle UI changes.

Repo-side runner:

node --import tsx scripts/harness/storyboard-transitions.ts

Installed-pack runner:

node ./.video-evaluator/agent/run-tool.mjs storyboard-transitions

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.