What problem does it solve? Managing generative media experiments is hard when results, parameters, and seeds are scattered. This Skill provides a structured workspace for creating reproducible procedural SVG artwork locally and submitting API workflows to an existing local ComfyUI instance, with every run recorded and inspectable. ## Core Features & Use Cases - Deterministic local generation: Run the studio toolchain's generate command to produce procedural SVG studies and a contact sheet, with hashes verifying reproducibility from palette and seed. - ComfyUI integration: Submit a saved API-format workflow (workflow.json) to a local ComfyUI over COMFYUI_URL, poll job history, and collect returned PNG outputs without model weights or paid APIs. - Run tracking and inspection: Successful artifacts and measurements live in out/runs/<id>/, out/latest.json names the current result, and failed runs preserve the last success with the error recorded in the verdict. - Use Case: Pick a palette and seed, generate a family of SVG studies, verify reproducibility via hashes, then submit the supplied EmptyImage/SaveImage workflow to your local ComfyUI and retain the returned PNGs for comparison. ## Quick Start Ask the agent to read studio.json, run the studio toolchain generate command, and report which run in out/runs contains the new result.
