AgentSkillsCN

fal-generate

利用 fal.ai 模型生成图像与视频。当用户要求创作、生成或制作图像、图片、视频、动画，或任何视觉内容时，可使用此功能。支持基于队列的文本转图像、文本转视频，以及图像转视频的生成过程。

SKILL.md
--- frontmatter
name: fal-generate
description: Generate images and videos using fal.ai models. Use when the user asks to create, generate, or make images, pictures, videos, animations, or visual content. Supports text-to-image, text-to-video, and image-to-video generation with queue-based processing.
metadata:
  author: falagent
  version: "0.1.0"

fal.ai Image & Video Generation

Generate images and videos from text prompts using fal.ai's AI models.

Setup

If FAL_KEY is not configured, run:

bash
cd /Users/yzlabmac/agentspace/falagent && bun run setup

Or set directly:

bash
export FAL_KEY=your_key_here

Generate Images

bash
cd /Users/yzlabmac/agentspace/falagent && bun run generate --prompt "DESCRIPTION" --model MODEL [--size square|portrait|landscape] [--num-images N]

Recommended image models:

  • •fal-ai/nano-banana-pro - Best quality, semantic-aware generation (default)
  • •fal-ai/flux-pro/v1.1-ultra - High quality, fast
  • •fal-ai/recraft-v3 - Stylized images

Generate Videos

bash
cd /Users/yzlabmac/agentspace/falagent && bun run generate --prompt "DESCRIPTION" --model MODEL

Recommended video models:

  • •fal-ai/veo3 - Highest quality text-to-video
  • •fal-ai/kling-video/v2.1/standard - Fast text-to-video
  • •fal-ai/minimax-video/video-01-live - Good for short clips

Image-to-Video

First upload the image, then generate:

bash
cd /Users/yzlabmac/agentspace/falagent && bun run upload --file /path/to/image.jpg
cd /Users/yzlabmac/agentspace/falagent && bun run generate --prompt "MOTION DESCRIPTION" --model fal-ai/kling-video/v2.1/pro/image-to-video --image-url IMAGE_URL

Recommended image-to-video models:

  • •fal-ai/kling-video/v2.1/pro/image-to-video - Best quality
  • •fal-ai/wan/v2.1/image-to-video - Good alternative

Async Mode (for long videos)

For models that take a long time, use --async to get a request ID immediately:

bash
cd /Users/yzlabmac/agentspace/falagent && bun run generate --prompt "..." --model fal-ai/veo3 --async

Then check status or get the result:

bash
cd /Users/yzlabmac/agentspace/falagent && bun run generate --status REQUEST_ID --model fal-ai/veo3
cd /Users/yzlabmac/agentspace/falagent && bun run generate --result REQUEST_ID --model fal-ai/veo3

Search Models

bash
cd /Users/yzlabmac/agentspace/falagent && bun run search --query "flux"
cd /Users/yzlabmac/agentspace/falagent && bun run search --query "video" --category text-to-video

Get Model Schema

bash
cd /Users/yzlabmac/agentspace/falagent && bun run scripts/get-schema.ts --model fal-ai/flux/dev

Output Format

All scripts output JSON to stdout and status messages to stderr. The JSON result from generate contains:

  • •For images: { "images": [{ "url": "https://...", "content_type": "image/jpeg" }] }
  • •For videos: { "video": { "url": "https://..." } }

Always show the user the URL(s) from the result so they can view the generated content.