canvas-cowork

Cowork on a spatial canvas from the CLI — create canvases, generate images/text/video/agent responses, read results, recall past work, and manage nodes on Flowith's shared canvas.

Your AI agent gets a live cursor on the canvas. It moves in real time — users see nodes appear and the tree grow.

npx skills add flowith-ai/canvas-cowork

• Create canvases and place nodes on a spatial workspace
• Generate images, text, video, and multi-step agent responses
• Read results back — images as markdown, text inline, video as links
• Recall past work across all canvases with semantic search
• Batch independent generations in parallel for speed

# Create a canvas and generate an image
bun $S --bot claude-code create-canvas "Dog Artwork"
bun $S --bot claude-code set-mode image
bun $S --bot claude-code submit "a golden retriever in a wheat field" --wait

# Burst: many items at once
bun $S --bot claude-code submit-batch "golden retriever" "husky" "corgi" "poodle"
bun $S --bot claude-code read-db --full

# Recall past work
bun $S recall "cyberpunk logo" --type image

| Mode | Use for | |------|---------| | text | Answers, writing, analysis | | image | Visual generation | | video | Video clips | | agent / neo | Multi-step research, planning, complex deliverables |

• A Flowith account with an active session in the browser
• Bun runtime

Full command reference and usage patterns are in SKILL.md.

MIT