Skip to content

nexu-io/openmotion / skills/openmotion-video-to-template

openmotion-video-to-template

Verified

Rebuild a reference video (pure motion graphics or mixed media with products, people, or audio) into a reusable editable OpenMotion template with source files, assets, preview, editing guide, and folder or ZIP delivery. Use only when a reusable template is requested; use openmotion-recreate for a single film recreation, openmotion-template-to-video for adapting an existing template, and openmotion-director for original videos.

FileKindSize
SKILL.mdskill md7.7 KB
agents/openai.yamlconfig146 B
references/measure.mdmarkdown2.4 KB
references/rebuild.mdmarkdown6.0 KB
references/rule-sources.mdmarkdown3.5 KB
references/verify.mdmarkdown8.6 KB

references/rule-sources.md

View on GitHub
# Guidance sources and limits

These sources explain guidance, not a guarantee of output quality. Methods remain the agent's choice within the requested result and native runtime contract. Human review decides visual acceptance.

## Material actually reviewed

- The pinned HyperFrames 0.9.7 authoring guide, format references and CLI help: native source structure, seeking, inspection, measurement and rendering. These establish runtime compatibility, not an exclusive reconstruction method.
- [Hypit public Skill](https://github.com/hypit-ai/hypit/blob/main/skills/hypit/SKILL.md) and [agent production guide](https://github.com/hypit-ai/hypit/blob/main/docs/guide/skill.md), reviewed 2026-11-04: persistent objects and changing relationships, grouping through shared behavior, parameterizing useful edits, keeping Skill/runtime/project roles separate. We adopt those decisions, not Hypit's project language or runtime.
- [Remotion Agent Skills documentation](https://www.remotion.dev/docs/ai/skills), reviewed 2026-11-04: separate relevant creation, preview, render and editing knowledge. This is documentation research, not a same-video effect comparison.

## Local observations

The 2026-11-04 controlled Grok test produced two different representations: independent SVG objects with GSAP animation, and a 373-frame contour transcription of about 6.7 MB. This supports inspecting whether intended edits remain practical. It does not prove every array, loop, animated path or per-frame renderer is unsuitable.

An earlier installed-package test created a Grok project in one session; a separate session received only its ZIP and an unchanged-render request. It actually re-rendered 373 frames at 30fps; decoded frames and MP4 bytes matched the delivered preview. A follow-up brand/image/long-text edit produced a decodable video but disconnected the headline's native variable binding. Source text stayed editable. These are single-sample observations, not proof for later Skill versions or other machines.

Earlier team tests, before 2026-11-04, recorded these failures. They explain the corresponding checks; they are sample observations, not universal effect guarantees:

- A team demo traced a 12-second film into about 2.8 MB of per-frame coordinates. Similarity was high, but shapes and motion could not be edited.
- Placing reference frames underneath the rebuild looked close while leaving nothing editable.
- A project that named only a system font changed text widths and wrapping on another machine.
- A Linear rebuild smoothed all three original hard cuts into continuous motion. The rhythm flattened while the similarity score stayed acceptable.
- A self-contained HTML omitted its audio file, and encoding failed with "file not found".
- An environment failure in a comparison tool was reported only as an error and was easily read as a pass.
- Creating sessions for Firecrawl and Vercel samples relabelled their own re-export as "independent session passed". The export matched, but the label was self-applied.

A copy rendered by its creator establishes self-containment, not independent-session or cross-machine reuse. A baseline must come from the delivered source. Missing assets, hidden dependencies and nondeterministic state can change a subsequent render; checks must report what actually ran.

Older unattributed claims about a competitor's five samples or file-hash practices are not evidence and are removed. No competitor is claimed to have better or worse effects without an actual comparison. Do not turn a single observed failure into a universal implementation restriction.
slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack."Stars: 198kUpdated Claude CodeCodexVerifiedInstall
ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report.Stars: 194kUpdated Claude CodeCodexVerifiedInstall
ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt.Stars: 194kUpdated Claude CodeCodexVerifiedInstall