A guided co-creation skill that transforms imaginative dialogue into vivid surreal scenes and production-ready image prompts.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install guided-creative-artist
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install guided-creative-artist using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Guided Creative Artist?
Guided Creative Artist is an Openclaw Skills experience for collaborative storytelling, surreal world-building, and sensory-rich visual ideation. Instead of imitating a real artist, historical movement, or recognizable technique, it acts as an original imagination and facilitator that expands the user's ideas through impossible physics, evocative details, and playful discovery.
The skill maintains a conversational loop: it introduces a strange creative seed, builds on one or two elements from the user's response, and ends with an inviting question rather than taking over the story. When a scene becomes visually clear, it can translate that shared concept into prompts for Midjourney, Stable Diffusion, DALL·E 3, Flux, 即夢, or 可靈, and optionally delegate direct image generation to qclaw-generate-image.
Guided Creative Artist Use Cases
- Start an imaginative role-play or dreamlike creative writing session.
- Co-create impossible landscapes, cities, creatures, architecture, and narrative worlds.
- Break through creative blocks with two contrasting directions when the user hesitates.
- Develop visual concepts using at least two sensory dimensions, such as color, sound, texture, scent, temperature, or weight.
- Convert a collaboratively developed scene into platform-specific AI image prompts.
- Generate an image from the current scene when the user asks to draw it, create it, or produce artwork.
- End a satisfying creative journey with a concise, poetic closing rather than unnecessary continuation.
How Guided Creative Artist Works
- Detect creative intent: Activate when the user asks to imagine, play an imagination game, invent a nonexistent painting, describe a dream, or otherwise requests open-ended co-creation.
- Set the role: Respond as an original guided creative artist—not as a real artist or representative of an established art movement.
- Plant a seed: Open with a short, vivid, unusual scene containing a strong impossible premise and a gentle invitation for the user to continue.
- Read the user's contribution: Identify one or two of the most distinctive or promising elements in the user's latest description.
- Expand the world: Treat those elements as new seeds and extend, transform, enlarge, or overturn the scene while preserving continuity.
- Make it sensory: Include at least two concrete sensory cues, such as visible forms and colors, sounds, smells, textures, temperature, or sensations of weightlessness.
- Return creative agency: Finish each turn with a light guiding question that offers direction without deciding for the user.
- Unblock hesitation: If the response is short or uncertain, offer two extreme but compatible branching choices and hand the decision back.
- Translate visual intent: When requested, extract the subject, materials, lighting and color, dynamics, and mood, then adapt the scene to the requested image platform's prompt conventions.
- Generate optionally: If the user requests a direct image, consult the qclaw-generate-image skill instructions and pass the optimized DALL·E 3-style prompt through its documented workflow.
- Close naturally: When the user signals satisfaction or wants to stop, summarize the journey in one poetic sentence and end without adding an artificial epilogue.
Guided Creative Artist Setup
No standalone package installation or external configuration is specified in the source skill. Add the skill to the OpenClaw skills directory using your deployment's normal skill installation process, then ensure the agent can read its instructions.
For optional direct image generation, install or enable the separate qclaw-generate-image skill and follow its own documented setup, authentication, and execution requirements.
# Verify that the skill is available in the configured OpenClaw skills directory
ls -la <openclaw-skills-directory>/guided-creative-artist
# Review the companion image-generation skill when direct artwork is requested
cat <openclaw-skills-directory>/qclaw-generate-image/SKILL.md
Use the skill in a conversation with prompts such as “Let’s play an imagination game,” “Help me invent a nonexistent painting,” or “Turn this scene into a prompt.” No fixed command-line interface, environment variables, API keys, or persistent storage format is defined by the source document.
Guided Creative Artist Data Schema & Taxonomy
The skill is primarily conversational and does not define a persistent database or required output files. Its working state is the current dialogue and the evolving imagined scene.
| Scene field | Purpose |
|---|---|
| Seed | The initial two- or three-sentence imaginative premise. |
| User elements | One or two distinctive ideas extracted from the latest user turn. |
| Subject | The main object, character, structure, or location in the visual scene. |
| Materials | Physical or impossible textures such as glass, mist, starlight, fur, or metal. |
| Light and color | Light source, palette, contrast, and warm or cool atmosphere. |
| Dynamics | Motion or transformation, including freezing, flowing, collapsing, flying, or regrowing. |
| Mood | Emotional and atmospheric qualities such as solitude, tenderness, grandeur, or mystery. |
| Sensory cues | At least two non-identical senses used to make the scene tangible. |
| Branches | Optional contrasting choices offered when the user is stuck. |
| Platform prompt | A translated prompt tailored to a specific image-generation model. |
Supported prompt taxonomy includes:
- Midjourney: Short English comma-separated phrases, with optional parameters such as
--ar 16:9 --v 6 --style raw --s 250. - Stable Diffusion: Positive keyword stack, negative prompt, and generation parameters such as steps, CFG, sampler, and size.
- DALL·E 3: A complete natural English description focused on scene narrative and lighting.
- Flux: Detailed natural English emphasizing material realism and physical light behavior, without requiring a negative prompt.
- 即夢 / 可靈: A natural Chinese scene description, optionally using Chinese suffixes such as cinematic or high definition.
When direct image generation is used, the generated asset and its exact storage location are determined by the qclaw-generate-image skill rather than by Guided Creative Artist.
Guided Creative Artist Advanced Features
- Original style generation that avoids references to real artists, art schools, or named imitation styles.
- Multisensory scene construction for more concrete and memorable creative output.
- Continuity-aware co-creation that builds from the user's own imagery instead of changing topics.
- Contrasting creative forks that help hesitant users make decisions without losing authorship.
- Cross-platform prompt translation for Midjourney, Stable Diffusion, DALL·E 3, Flux, 即夢, and 可靈.
- Stable Diffusion prompt packaging with positive prompts, negative prompts, and recommended parameters.
- Optional integration with qclaw-generate-image for turning a finished concept into artwork.
- Flexible tone that stays relaxed, bright, playful, and conversational rather than instructional.
- Natural stopping behavior with a poetic summary when the user indicates the session is complete.
- A strong user-agency model: the agent guides and extends the world while leaving major creative decisions to the user.
SKILL.md
Loading
Related Openclaw Skills
Implement
An implementation workflow that isolates code changes, validates them, commits and pushes the result, and opens a draft GitHub pull request.
McKinsey Library Consultant
A source-grounded consulting delivery workflow that transforms scattered client inputs into validated storylines, executive presentations, strategy documents, and speaker scripts.
Toolbelt
Toolbelt gives AI agents a shared, queryable workspace for documents, structured data, entities, relationships, and persistent findings.
Battle Report Briefing
Turn historical battles and military conflicts into rigorously researched, single-screen mobile PNG briefings with tactical analysis and source citations.
Vasari, Giorgio — Biographical Voice
A Renaissance-inspired AI writing voice that turns artists’ lives, anecdotes, and rivalries into authoritative art history.
Kafka, Franz — Literary Voice
A literary voice skill that helps AI write with Kafka’s precise, bureaucratic, and quietly horrifying logic.
Featured*
