Traction scoreGitHub stars can be faked, so popularity alone can be misleading. Traction Score looks for broader signs of recent attention, adoption, and active maintenance.
This skill collection bridges AI ad generation and Meta ad deployment. Using the Arcads external API, it produces video and image assets across models like Seedance 2.0, Sora 2, and Veo 3.1, then feeds finished creatives into the Meta Marketing API as paused ads in existing ad sets. Supporting workflows include competitor research, copy modeling from winning ads, and reverse-engineering reference videos into reusable templates. It serves performance marketers, creative operations teams, and DTC brands who need to move from a generated concept to a published ad within the agent.
WHO IT'S FOR
Performance marketers running ad creative pipelines
deploy AI-generated creative as live ads
Creative ops teams iterating AI video ads
generate and clone video ads via external api
DTC brand content strategists
build reusable ad templates from competitor analysis
Social media ad buyers deploying platform-native assets
Compatible AgentsThe repository documents support for these agents. The skills may also work with other agents that can load SKILL.md files, but they may need some setup or small changes.
Follow the documented setup, then try a first task.
I’m using
Claude Code bundle
Installs: Complete Arcads AI Video Agent Skill Pack repository; the listed skillPaths are a bounded selection of positively documented skills, not the bundle's complete inventory. · Claude Code
Before you start
Python 3.10+
An Arcads account and Arcads API key
bash-compatible environment; Windows users are advised to use WSL2
ffmpeg for the analyze-video starter task
Optional openai-whisper Python package for automatic transcription; dialogue can instead be supplied manually
Platform: macOS or Linux with bash; WSL2 recommended on Windows
In a terminal, start in the directory where you want the repository created. Clone it and enter the repository directory; keep the same terminal session for the next step.
git clone https://github.com/krusemediallc/arcads-claude-code.git
cd arcads-claude-code
From the cloned repository directory, run the setup script in the terminal. When prompted, provide the Arcads API key from the account API settings; the script creates .env, verifies the connection, creates MASTER_CONTEXT.md, and synchronizes the skills.
./scripts/setup.sh
Open the cloned repository folder in Claude Code. Its SessionStart hook reports the installed skills and setup state.
Installs: Complete Arcads AI Video Agent Skill Pack repository; the listed skillPaths are a bounded selection of positively documented skills, not the bundle's complete inventory. · Cursor
Before you start
Python 3.10+
An Arcads account and Arcads API key
bash-compatible environment; Windows users are advised to use WSL2
ffmpeg for the analyze-video starter task
Optional openai-whisper Python package for automatic transcription; dialogue can instead be supplied manually
Platform: macOS or Linux with bash; WSL2 recommended on Windows
In a terminal, start in the directory where you want the repository created. Clone it and enter the repository directory; keep the same terminal session for the next step.
git clone https://github.com/krusemediallc/arcads-claude-code.git
cd arcads-claude-code
From the cloned repository directory, run the setup script in the terminal. When prompted, provide the Arcads API key from the account API settings; the script creates .env, verifies the connection, creates MASTER_CONTEXT.md, and synchronizes the skills.
./scripts/setup.sh
Open the cloned repository folder in Cursor. The synchronized skills are exposed under .cursor/skills/.
Creative direction describing the offer, audience, headline, and visual treatment
Aspect ratio: 1:1
Optional local reference-image paths
Configured ARCADS_BASIC_AUTH or ARCADS_API_KEY
Use the chatgpt-image-ad skill to create one standalone 1:1 Meta image-ad creative with gpt-image-2. My creative direction is [describe the offer, audience, headline, and desired visual treatment]. Use these optional local reference-image paths: [paths, or say none]. Estimate the Arcads credit cost and ask for confirmation before generating. Generate one variant only and do not upload it to Meta.
Local reference folder containing any required face, logo, product, example, or style images
Configured Arcads credentials and productId
Use the generate-youtube-thumbnail skill to create one thumbnail concept for a video about [video topic]. The subject is [person, AI character, or no person], the thumbnail text is [short text], and the desired brand colors are [colors]. Use the actual local files in [reference folder path] for any face, logo, or product that must match. First verify that the files exist, then estimate the cost and ask for confirmation before generating one variation.
Turn a reference video into a reusable Seedance template
Uses analyze-video
Local path to an MP4, MOV, WEBM, or similar video
Optional style name
Installed ffmpeg/ffprobe
Optional Whisper installation, or manually supplied dialogue
Use the analyze-video skill to analyze the local video at [path/to/reference-video.mp4] and create a reusable Seedance 2.0 prompting template named [optional style name]. Distill the style into a single 15-second format with no more than three beats and two or three spoken lines, and include a short multi-clip strategy if the source is longer. ffmpeg is installed at [path or available on PATH]. Use Whisper if available; otherwise ask me for the dialogue.
Claude
Cursor
Gemini CLI
Copilot
Grok