Capabilities
The Virse MCP Skill handles everything from simple one-shot commands to multi-step creative workflows. Just describe what you want in natural language — your MCP client takes care of the rest.
Simple Commands
These are direct actions that map to a single tool call:
"Show my workspaces"
Lists all your workspaces
"What's on my canvas?"
Shows canvas contents
"Check my balance"
Displays account and CU balance
"Search for sunset photos"
Searches your image assets
"What models are available?"
Lists all AI image models
"Show my asset folders"
Lists your organized folders
"Details of element abc1"
Inspects a specific canvas element
"Trace connections from abc1"
Maps the connection graph from an element
"What prompt was used for this?"
Shows generation details for an asset
"Create a new workspace"
Creates a workspace
"Generate an image of X"
Generates a single image
"Add this image to folder Y"
Links an image to a folder
"Upload this image"
Uploads a local image
Complex Workflows
For multi-step tasks, the agent orchestrates multiple tool calls using built-in playbooks:
"Generate 10 variations of this concept"
Batch image generation with layout
"Summarize what's on this canvas"
Full workspace overview and analysis
"Build a moodboard from references"
Searches and assembles a reference board
"Clean up my canvas"
Finds orphaned elements, organizes layout
"Compare these models side by side"
Generates the same prompt across models
"Collect assets from all workspaces"
Cross-workspace asset consolidation
"Organize my assets into folders"
Curates and categorizes assets
"Refine this prompt"
Iterative prompt improvement with feedback
"How was this image created?"
Traces the full creation lineage
"Analyze the graph structure"
Maps canvas topology and node relationships
Model Selection
The agent can recommend the best model for your task. Here's a quick guide:
Nano Banana 2
Fast, cheap default for quick iterations
Gemini 3 Pro
High-quality output
Imagen 4.0 Ultra
High-quality output
GPT Image 1.5
Text rendering and complex instructions
FLUX Kontext Pro / Max
Style transfer and reference-based generation
FLUX 1.1 Pro Ultra
Precise resolution control
More models may be available — ask the agent to list current models for the latest.
Prompt Tips
Get better results by being specific with your prompts:
Front-load the subject — Put the main subject first in your prompt
Add style details — Include lighting, composition, and medium: "cinematic lighting", "flat vector illustration", "35mm photography"
Be specific — "Golden hour sunset over snow-capped mountains" beats "sunset"
Use references with FLUX Kontext — Provide a source image and describe the transformation you want
Multi-image references — You can pass up to 10 reference images for a single generation
Aspect Ratios
When generating images, the agent automatically sizes elements to match your chosen aspect ratio:
1:1 (default)
512
512
16:9
512
288
9:16
288
512
4:3
512
384
3:4
384
512
Last updated