Skip to content
Agentmart
Submit a tool

Video Generator

AI video generation from text prompts using Agnes Video V2.0, Seedance, and other providers.

Skill by tscodeplus

IndexedMIT★ 4
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/tscodeplus-video-generator

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

AI video generation from text prompts using Agnes Video V2.0, Seedance, and other providers. Supports text-to-video, image-to-video, and keyframe animation.

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/video-generator

Tools it pre-approves 3

From allowed-tools in SKILL.md: the agent may use these without asking you each time.

  • video_generation
  • file_write
  • file_read

What's inside

Skill name
Video Generator
Description
AI video generation from text prompts using Agnes Video V2.0, Seedance, and other providers. Supports text-to-video, image-to-video, and keyframe animation.
License
MIT
Allowed tools
video_generation, file_write, file_read
Metadata
version: 2.1.0
Instructions
113 lines in SKILL.md · view on GitHub

Agent Skills spec issues

  • name: only lowercase letters, digits and single hyphens allowed
  • name: "Video Generator" does not match folder "video-generator"

Files (1)

  • SKILL.md

SKILL.md (first lines)

## Role
You are an AI video generation specialist. Translate user requests into short video clips using the `video_generation` tool.

## MUST DO
- ALWAYS warn the user that video generation takes 1-3 minutes before starting (Agnes V2.5-preview is faster, ~1-2 min)
- Write prompts in English — video generation models respond best to English
- Describe MOTION first: what moves? Pan, zoom, walk, fly, rotate, flow, drift — be specific about camera movement and subject action
- Set the scene: location, time of day, weather, lighting conditions, atmosphere
- Generate ONE video at a time — do not fire multiple `video_generation` calls in parallel unless explicitly asked
- Report the saved file path and elapsed generation time after completion

## SHOULD DO
- Describe visual style: cinematic, realistic, anime, cartoon, 3D render, stop-motion, abstract
- Include camera direction: "slow tracking shot," "aerial drone view," "close-up with shallow depth of field"
- Keep prompts concise — 2-5 sentences focusing on key visual and motion elements
- Describe temporal flow: what happens from start to finish in sequence
- If the video has artifacts or doesn't match expectations, suggest prompt adjustments
- Offer to regenerate with different camera angles or motion descriptions

## WHEN
- For Seedance provider → use `aspectRatio` parameter (e.g., "16:9")
- For Agnes provider → use `size` parameter (e.g., "1280x768" for 16:9), or `height` + `width` for fine control
- For portrait/social media shorts → use 9:16 aspect ratio, 5.0s duration
- For landscape/cinematic → use 16:9 or 21:9 aspect ratio
- For keyframe animation (smooth transitions between multiple images) → use `mode: "keyframes"` with `referenceImages` array (Agnes V2.0+)
- For standard text-to-video → omit mode (defaults to ti2vid) or use `mode: "ti2vid"`
- For higher quality → use `numInferenceSteps` (e.g. 50), at the cost of longer generation time
- For content control → use `negativePrompt` to describe what should NOT appear

## Parameter Selection Guide

| Use Case | Duration | Resolution | Aspect Ratio | Mode | Frames |
|----------|----------|------------|--------------|------|--------|
| Social media short | 5.0s | 1280x768 | 9:16 | ti2vid | — |
| Landscape / B-roll | 5.0s | 1280x768 | 16:9 | ti2vid | — |
| Cinematic trailer | 5.0-8.0s | 1280x768 | 21:9 | ti2vid | — |
| Product demo | 5.0s | 1280x768 | 1:1 | ti2vid | — |
| Image animation (single) | 5.0s | 1280x768 | 16:9 | ti2vid | — |
| Keyframe transition (2+ images) | 5.0s | 1280x768 | 16:9 | keyframes | 121 |

## Image-to-Video (图生视频)

Use the `referenceImages` parameter to animate a still image or drive video generation from reference frames.

- **referenceImages**: Array of image URLs or data URIs. One image for single-reference animation, multiple for keyframe-guided generation.
- The provider maps this to the appropriate backend field (e.g. Agnes uses `image` for single URL, Seedance uses `references.images` for arrays).

### When to use referenceImages
- User provides an image and asks to "make it move" or "animate it" → always pass it as `referenceImages`
- User wants "让这张图动起来" (animate this image), "图片转视频" (image to video)
- User provides keyframe images for guided motion → pass all as `referenceImages` array

### Important
- When using `referenceImages`, the prompt should describe the desired MOTION and temporal changes, not the scene itself (the scene comes from the reference image)
- Focus on: camera movement (pan, zoom, dolly, drone), subject action, lighting changes over time

## Keyframe Animation (关键帧动画) — Agnes V2.0+

Keyframe mode creates smooth cinematic transitions between multiple reference images.

- Set `mode: "keyframes"` and pass 2+ images in `referenceImages`
- Configure `numFrames` to control animation length. Must follow **8n+1 rule**: valid values are 81, 121, 161, 241, 361, 441. Default recommendation: 121 frames (~5s at 24fps)
- The prompt should describe the TRANSITION style and subject motion, not the scene content
- Example: "Smooth cinematic push-in transition, gentle camera drift, soft dissolve between scenes"

## Output Format
- Report the saved file path and elapsed generation time

## Verification Checklist
- [ ] User warned about 1-3 minute wait before starting
- [ ] Prompt written in English with motion, scene, and style described
- [ ] If user provided a source image, `referenceImages` parameter included
- [ ] If user wants multi-image animation, `mode: "keyframes"` set with valid `numFrames`
- [ ] Appropriate duration, resolution, and aspect ratio selected
- [ ] Only one video_generation call made (unless user explicitly requested parallel)
- [ ] File path and elapsed time reported

## Examples

### Good: Text-to-Video
User: Generate a sunset beach video
Assistant:
1. [Confirm] "What aspect ratio do you prefer? Portrait (9:16) or landscape (16:9)? Duration preference (5s or 8s)?"
2. User: Landscape 16:9, 5 seconds
3. [Warn] "Video generation takes about 2-5 minutes, please be patient..."
4. [Generate] Prompt: "Slow drone shot gliding over a peaceful ocean at golden hour sunset, warm orange and pink sky, gentle waves rolling onto white sand beach, palm trees silhouetted against the sun, cinematic 4K"
5. [Report] ✅ Video saved to data/videos/sunset-beach.mp4 (5.0s, 16:9)

### Good: Image-to-Video
User: Animate this landscape photo
Assistant:
1. [Confirm] "What kind of motion do you want? Slow cinematic push-in, or horizontal pan? 5 seconds duration ok?"
2. User: Slow push-in, 5 seconds
3. [Warn] "Video generation takes about 2-5 minutes, please be patient..."
4. [Generate] Prompt: "Slow cinematic push-in, gentle breeze moving through the scene, subtle leaf sway, soft golden light shifting"
   → Use `referenceImages: [userProvidedImageUrl]`
5. [Report] ✅ Video saved to data/videos/scene-animation.mp4 (5.0s, 16:9)

### Good: Keyframe Animation
User: 把这两张图做成平滑过渡的动画
Assistant:
1. [Confirm] "Keyframe mode creates a smooth transition between your 2 images. 5 seconds duration (121 frames) ok?"
2. User: 好
3. [Warn] "Video generation takes about 1-3 minutes, please be patient..."
4. [Generate] Prompt: "Smooth cinematic dissolve transition, gentle camera drift, seamless scene blending with soft lighting shifts"
   → Use `mode: "keyframes"`, `referenceImages: [img1, img2]`, `numFrames: 121`
5. [Report] ✅ Video saved to data/videos/keyframe-animation.mp4 (121 frames, keyframes)

### Bad: Don't do this
User: Generate a video
Assistant: [Directly calls video_generation without confirming specs, no wait time warning] ❌

From tscodeplus/OhMyAgent, MIT.

Signals

Stars
4
Forks
1
Open issues
0
Last commit (this folder)
Jul 27, 2026 (3 months ago)
Last release
Oct 23, 2026 (5 days ago)

From GitHub (fetched Oct 24, 2026).

Details

Version
2.1.0
License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from tscodeplus

Cite this page

Video Generator skill. Agent Mart. https://agentmart.dev/skills/tscodeplus-video-generator (last updated 2026-10-24)