菜单
Autonomously creates a py5 media art still image following the workflow in AGENTS.md. Covers branch setup, past-work research, implementation, commit, and push. Triggers: create artwork, new sketch, media art
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Analyzes a py5 media art sketch and generates a Medium-ready technical article in Markdown. Output goes to medium/YYYY-MM-DD_{work_name}_medium.md. Triggers: write medium article, blog post, medium draft, write article about
Re-encodes an existing py5 artwork's MP4 to mobile-optimized format: H.264/10–20s seamless loop, two orientations (landscape 1280×720 and portrait 720×1280). Takes a work name as argument. Triggers: export mobile, mobile export, mobile video, re-encode for mobile
Artist agent for py5 generative media art. Responsible for concept design, implementation, and revisions based on critic feedback.
Creates exactly one py5 media art still image with a lightweight workflow for slow Claude Code proxy routes such as NVIDIA API. Minimizes planning/review overhead while still generating preview images, updating records, committing, and pushing. Triggers: create artwork lite, lightweight artwork, NVIDIA proxy artwork
Continuously creates py5 media art still images with the lightweight workflow for slow Claude Code proxy routes such as NVIDIA API. Repeats create-artwork-lite-style iterations until interrupted, committing and pushing each work. Triggers: create artworks lite, batch artwork lite, continuous lightweight artwork, NVIDIA proxy artworks
Continuously creates py5 media art still images in a loop. Each iteration follows the same workflow as create-artwork (plan → implement → review → commit). Runs until stopped. Triggers: create artworks, batch art, continuous sketches
| name | create-artwork |
| description | Autonomously creates a py5 media art still image following the workflow in AGENTS.md. Covers branch setup, past-work research, implementation, commit, and push. Triggers: create artwork, new sketch, media art |
| allowed-tools | Bash, Read, Write, Edit |
Autonomously creates a py5 media art still image by following the workflow defined in AGENTS.md.
AGENTS.md, .agents/skills/shared/artwork-conventions.md, and .agents/skills/shared/py5-templates.md to confirm the full workflow and shared conventionssketch/WORKS.md and .agents/FEEDBACK.md, then use the logic-lab MCP server to search for relevant algorithm references before producing a Creative Brief with Format: Still image (see .agents/skills/planner/SKILL.md)feature/works-YYYYMMDDsketch/{work_name}/main.py; if the brief includes a Logic Lab Reference, fetch only those referenced files through the logic-lab MCP server (see .agents/skills/artist/SKILL.md).agents/skills/critic/SKILL.md)sketch/WORKS.md and .agents/FEEDBACK.md (leave Rating and Comment empty per conventions).agents/skills/shared/artwork-conventions.mdmain.pylogic-lab MCP server for external algorithm references. If the MCP server is unavailable, continue without logic-lab rather than relying on copied indexes from another repository.logic-lab. Search first, then fetch only the specific algorithm files named in the Creative Brief.