Fal.ai API — 技能 — openclaw中文资讯站

技能详情（站内镜像，无评论）

Generate images, videos, and audio via fal.ai API (FLUX, SDXL, Whisper, etc.)

媒体与内容

许可证：MIT-0

MIT-0 ·免费使用、修改和重新分发。无需归因。

版本：v0.1.0

统计：⭐ 1 · 3.3k · 18 current installs · 18 all-time installs

⭐ 1

安装量（当前） 18

🛡 VirusTotal ：良性 · OpenClaw ：良性

Package：agmmnn/fal-ai

安全扫描（ClawHub）

VirusTotal ：良性
OpenClaw ：良性

OpenClaw 评估

The skill's code, documentation, and requested environment variable (FAL_KEY) are consistent with its stated purpose of calling fal.ai's queue API to generate media; nothing in the package appears disproportionate or unrelated.

目的

The name/description (fal.ai media generation) matches the code and SKILL.md: the package implements HTTP calls to a queue endpoint for image/video/audio models and only needs an API key (FAL_KEY). Model mappings, endpoints, and usage examples are consistent with the stated capability.

说明范围

SKILL.md instructs the agent to set FAL_KEY (or clawdbot config) and to call the included Python client. The runtime instructions do not request unrelated files, broad system access, or unexplained data collection. Behavior is limited to submitting jobs, polling status, and returning results.

安装机制

There is no install spec and the code uses only the Python standard library (urllib, subprocess, etc.), so nothing arbitrary is downloaded or written to disk during install.

证书

The only required credential is FAL_KEY (declared as primary). As a fallback the client will call the 'clawdbot config' CLI via subprocess to read skill.fal_api.key — this is proportional to the stated setup but does mean the skill will invoke a local CLI if the env var is not set.

持久

The skill does not request always:true and does not modify other skills or system-wide settings. It runs when invoked and requires no elevated or persistent privileges.

综合结论

This package appears coherent: it only needs your FAL_KEY and uses a small Python client to call a queue endpoint (queue.fal.run). Before installing, confirm you trust the source (owner slug differs from the README image host), verify that queue.fal.run is the official fal.ai endpoint you expect, and avoid pasting a high-privilege key if you don't trust the skill. If you prefer, use the official fal.ai client (if available) or issue a scoped A…

安装（复制给龙虾 AI）

将下方整段复制到龙虾中文库对话中，由龙虾按 SKILL.md 完成安装。

请把本段交给龙虾中文库（龙虾 AI）执行：为本机安装 OpenClaw 技能「Fal.ai API」。简介：Generate images, videos, and audio via fal.ai API (FLUX, SDXL, Whisper, etc.)。
请 fetch 以下地址读取 SKILL.md 并按文档完成安装：https://raw.githubusercontent.com/openclaw/skills/refs/heads/main/skills/agmmnn/fal-ai/SKILL.md
（来源：yingzhi8.cn 技能库）

SKILL.md

打开原始 SKILL.md（GitHub raw）

---
name: fal-api
description: Generate images, videos, and audio via fal.ai API (FLUX, SDXL, Whisper, etc.)
version: 0.1.0
metadata:
  {
    "openclaw": { "requires": { "env": ["FAL_KEY"] }, "primaryEnv": "FAL_KEY" },
  }
---

# fal.ai API Skill

Generate images, videos, and transcripts using fal.ai's API with support for FLUX, Stable Diffusion, Whisper, and more.

## Features

- Queue-based async generation (submit → poll → result)
- Support for 600+ AI models
- Image generation (FLUX, SDXL, Recraft)
- Video generation (MiniMax, WAN)
- Speech-to-text (Whisper)
- Stdlib-only dependencies (no `fal_client` required)

## Setup

1. Get your API key from https://fal.ai/dashboard/keys
2. Configure with:

```bash
export FAL_KEY="your-api-key"
```

Or via clawdbot config:

```bash
clawdbot config set skill.fal_api.key YOUR_API_KEY
```

## Usage

### Interactive Mode

```
You: Generate a cyberpunk cityscape with FLUX
Klawf: Creates the image and returns the URL
```

### Python Script

```python
from fal_api import FalAPI

api = FalAPI()

# Generate and wait
urls = api.generate_and_wait(
    prompt="A serene Japanese garden",
    model="flux-dev"
)
print(urls)
```

### Available Models

| Model         | Endpoint                              | Type         |
| ------------- | ------------------------------------- | ------------ |
| flux-schnell  | `fal-ai/flux/schnell`                 | Image (fast) |
| flux-dev      | `fal-ai/flux/dev`                     | Image        |
| flux-pro      | `fal-ai/flux-pro/v1.1-ultra`          | Image (2K)   |
| fast-sdxl     | `fal-ai/fast-sdxl`                    | Image        |
| recraft-v3    | `fal-ai/recraft-v3`                   | Image        |
| sd35-large    | `fal-ai/stable-diffusion-v35-large`   | Image        |
| minimax-video | `fal-ai/minimax-video/image-to-video` | Video        |
| wan-video     | `fal-ai/wan/v2.1/1.3b/text-to-video`  | Video        |
| whisper       | `fal-ai/whisper`                      | Audio        |

For the full list, run:

```bash
python3 fal_api.py --list-models
```

## Parameters

| Parameter  | Type | Default          | Description                                        |
| ---------- | ---- | ---------------- | -------------------------------------------------- |
| prompt     | str  | required         | Image/video description                            |
| model      | str  | "flux-dev"       | Model name from table above                        |
| image_size | str  | "landscape_16_9" | Preset: square, portrait_4_3, landscape_16_9, etc. |
| num_images | int  | 1                | Number of images to generate                       |
| seed       | int  | None             | Random seed for reproducibility                    |

## Credits

Built following the krea-api skill pattern. Uses fal.ai's queue-based API for reliable async generation.