Which HyperFrames Skill Fits Your Video Edit?

Our selection rule is to name the failing layer first: motion, mix or overall story. Adding a larger orchestrator to a small edit can create unnecessary planning and dependencies. Conversely, asking a motion specialist to invent a complete product story leaves essential decisions unstated.

This Skill describes repeatable visual animation inside a HyperFrames composition, including zooms and authored reframing. It does not provide automatic face tracking, choose matching shots or own temporal clip editing. The source calls for motion diagnostics and inspection of meaningful frames.

Illustrative request—not an executed example: ‘Keep this shot and copy unchanged; zoom toward the feature panel, then hold the final view. Show the opening, peak zoom and final frame.’ Our acceptance criterion would be readable text and continuity of the same subject, not simply a successful command exit.

This package works on audio already placed in a composition: levels, fades, effects and automation. It distinguishes mixing from finding music or generating narration, and does not promise automatic waveform synchronization or drift correction.

Illustrative request: ‘Make the existing narration understandable over the existing music without changing either clip’s timing.’ We would keep both tracks, compare speech-heavy and quiet passages, and listen to the rendered output. Without listening or an appropriate audio check, we would report the mix as unverified rather than call it improved.

general-video is the broader composition workflow when a specialized route does not fit. It starts from existing project decisions when available and depends on other HyperFrames instructions for media, composition and motion. It is not a reason to restart a settled brief for every edit.

product-launch-video is the promotional route: product capture or supplied material, story planning, design, audio when needed, frame assembly and final review. Authentication and available local alternatives are separate considerations. A failed product capture is not permission to invent product screens or claims.

Our practical distinction: for ‘shorten this existing customer interview,’ first preserve the project and constrain the edit. For ‘explain our new feature to buyers,’ first approve the message, actual product evidence and delivery format. These are proposed briefs, not videos produced in this review.

Installing instruction files does not make every downstream service free. Review the selected media provider’s current charges, authentication and data handling before sending a brief or asset. Agree on the number of drafts and a spending ceiling; never place credentials in a public project or prompt.

Use only authorized product screens, voices, music and customer material. Treat capture, external media requests, local file changes and final rendering as distinct actions. Define acceptance before work: target format, readable frames, correct product claims, intelligible audio when present and a final artifact the user has actually reviewed.

• State whether this is an edit or a fresh creation.
• Identify required companion Skills and local tools before installing.
• Approve external data sharing and paid operations separately.
• Keep a source revision and inspect changes before updating.

We inspected the four linked SKILL.md files at heygen-com/hyperframes revision 58c873d8161d6a2357eacf89a540bfecc359349b on October 19, 2026. The public evidence record contains exact paths and SHA-256 hashes. It records file existence and identity, not a full dependency audit, installation, runtime compatibility, output quality or security certification. Current repository instructions may differ.

No video, audio, provider login, render or paid API request was executed for this article. We have no comparative speed, quality or conversion measurements. The proposed requests and acceptance checks are editorial examples, not user testimonials. The dated Top 100 snapshot is unchanged.