- Home
- Skills
- Image & Video
- Keyframe Video Production — multi-shot AI video from generat
Keyframe Video Production — multi-shot AI video from generat
A production workflow for keyframe-driven AI video, written from real runs in September–October 2026 and verified against a keyframe (first-frame) video model.
Free
Works with the AI Agents you already use
About this skill
A production workflow for keyframe-driven AI video, written from real runs in September–October 2026 and verified against a keyframe (first-frame) video model.
What it covers, gate by gate:
(1) Source analysis and storyboard — shot shape, shot mix, and what stays constant across the cut;
(2) Blocking — placing the film in a space before writing any shot;
(3) Image prompts — including a ten-property shot-signature check so a cut does not repeat itself;
(4) Character cards — one card per recurring character and per state, face taken from the close-up only;
(5) Keyframe generation and review — consistency, faces and mouths, framing variety, uncanny-valley checks;
(6) Temporary public hosting for first-frame URLs, with credential and metadata scans before upload;
(7) Video submission and assembly — dialogue, lip-sync, voice bibles, subtitles overlaid locally, and a clean master.
Hard rules built in:
- Every gate is a stop. Approval of one gate is never approval of the next.
- Hosting is its own approval, separate from approving the images.
- The agent cannot hear: audio analysis never proves language or lines; a human verdict does.
- Never state billing you did not check.
- No pure text-to-video, no editing existing footage, and nothing published outward without a fresh confirmation.
Verified capability notes (dated, not guarantees):
- Intelligible spoken dialogue with matching lip movement, directly from a keyframe and a prompt containing the line;
- Two distinct voices holding across one cut, each with its own voice bible;
- Voice-sample mode and its trade-off (reference mode takes audio but not a first frame);
- The 'one single unbroken take' sentence that stopped unscripted internal cuts, and when to replace it for scripted multi-shot clips;
- Why off-screen narration works in a probe and fails in production, and the visible-storyteller route that fixes it.
Who it is for: creators and agents making short films, episodes, demos or adapted works where character and scene continuity across shots matters.
Requirements: an agent that can read files, run shell commands and make HTTP calls (Claude Code, Codex, Hermes and similar); a keyframe/first-frame video model; an image model for keyframes. Video and image generation costs are the buyer's own.
Included: SKILL.md (English) and SKILL.zh.md (Chinese), same workflow in both languages.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Video Skills.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 18 days ago
- Free with an Agensi account
- Software
- Terminal accessKeyframeorfirst-frame video modelImage model
Your agent must also be able to make HTTP calls. The image model creates the keyframes.