Creative Studio

An agent skill for making social graphics, carousels, voiceovers, animation, and edited videos with VoiceStudio and the six ArtCraft creative apps.

Creative Studio teaches an agent how to choose the right app, inspect its live tools, save editable projects, and move assets between apps. It includes local Python helpers for MCP capability discovery and saving narration as WAV.

App	Creative role
VoiceStudio	Voiceovers, authorized voice cloning, transcription
PhotoCraft	Layered graphics, masks, thumbnails, post images
FilmCraft	Video assembly, cuts, reframing, audio, captions, final export
EffectCraft	Animated titles, keyframes, compositing, motion graphics
VectorCraft	Logos, icons, diagrams, scalable illustration
DesignCraft	Carousels, typography, coordinated page layouts
LightCraft	Photo/RAW development and consistent photo sets

Install the skill

For agents supported by the skills CLI:

npx skills add odomcl22/creative-studio-skill --skill creative-studio

To inspect available skills before installation:

npx skills add odomcl22/creative-studio-skill --list

Alternatively, clone this repository and copy skills/creative-studio into your agent's personal skill directory, such as ~/.codex/skills/creative-studio or ~/.claude/skills/creative-studio. Inspect an existing copy before replacing it.

For Claude's in-app skill library, download the ZIP from the latest GitHub release, then upload it in Customize → Skills. The ZIP contains the complete skill folder. See Claude's official instructions.

First use and installation choices

The skill performs read-only checks for ArtCraft Launcher, each of the six apps/helpers, and VoiceStudio. If additions are needed, the user gets independent Yes / No choices for ArtCraft and VoiceStudio, and can approve only selected ArtCraft apps. It provides official installation routes and respects refusals. It asks separately before an unapproved model pack or app-specific security exception.

The package contains instructions and helper scripts. Native apps, their MCP servers, runtime dependencies, and voice models are separate installations. Working local tools are needed for execution; loading the skill into a cloud chat does not grant access to a desktop's files or localhost.

Official apps: ArtCraft Launcher and VoiceStudio. Their licenses and model licenses remain separate from this skill's license.

Try it

• “Use Creative Studio to make a five-card carousel with consistent typography.”
• “Use Creative Studio to create a narrated vertical video with animated titles.”
• “Develop these photos in LightCraft and turn the strongest one into a PhotoCraft thumbnail.”
• “Make reusable vector icons, animate them in EffectCraft, and assemble the final video in FilmCraft.”

The skill uses only the apps that improve the deliverable. It preserves editable native projects, checks actual exports, and treats social publishing as a separately authorized account action.

Optional local helpers

Both helpers need Python 3.11+ and the packages in skills/creative-studio/requirements.txt. Reuse a compatible existing environment, including VoiceStudio's managed Python if available. To create a separate helper environment after approving its installation:

python3 -m venv .venv
.venv/bin/python -m pip install -r skills/creative-studio/requirements.txt

On Windows use .venv\Scripts\python.exe in place of .venv/bin/python.

Discover local ArtCraft MCP tools, supplying the actual helper directory:

.venv/bin/python skills/creative-studio/scripts/studio_tools.py \
  --app all --bin-dir /absolute/path/to/artcraft/helpers \
  --out capabilities.json

You can also put helpers on PATH or set ARTCRAFT_BIN_DIR. Probe one command without editing a document:

.venv/bin/python skills/creative-studio/scripts/studio_tools.py \
  --app filmcraft --commands timeline.place

Save a VoiceStudio narration without printing base64 audio into the conversation:

.venv/bin/python skills/creative-studio/scripts/render_voice.py \
  --text-file script.txt --out narration-v01.wav --language en

Add --voice <saved-profile-id> for an existing saved voice. Both helpers accept --port when VoiceStudio's local backend uses a port other than 3900. The narration helper refuses to overwrite a file. Check the generated audio for pronunciation and pacing before assembling a video.

Compatibility and validation

The initial app references were checked on macOS Apple Silicon against PhotoCraft 0.7.0, FilmCraft 0.6.0, EffectCraft 0.8.0, VectorCraft 0.9.0, DesignCraft 0.6.0, LightCraft 0.6.0, and VoiceStudio 0.5.6. The helper dependencies were tested with MCP SDK 1.29.0 and httpx 0.28.1. Other app versions and platforms should be checked using their actual tool schemas; full cross-platform app behavior is not claimed.

Live discovery, native-project saves for all six ArtCraft apps, representative exports, and a generated WAV were exercised during development. These apps evolve quickly; the skill requires capability discovery rather than assuming complete Adobe feature parity.

Package a release

python3 scripts/package_skill.py --version 1.0.0

This creates the distributable skill ZIP and SHA-256 checksum in dist/. See publication guidance for skills.sh, SkillsMP, ClawHub, and Hugging Face.

License and credits

MIT license for this skill and its original helper scripts. Created by Christian Lee Odom. The underlying applications and models retain their own licenses. This is an independent community workflow and is not affiliated with or endorsed by the app authors.