Fal.ai API

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

Security Scan

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.

Purpose & Capability

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.

Instruction Scope

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.

Install Mechanism

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.

Credentials

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.

Persistence & Privilege

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.

Assessment

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 API key that can be revoked. Note the client will call 'clawdbot config get ...' as a fallback if FAL_KEY is not set in the environment — if you don't run clawdbot or prefer not to expose that config, ensure FAL_KEY is set in the environment instead.

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:

export FAL_KEY="your-api-key"

Or via clawdbot config:

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

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:

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.