菜单
Start the chess server, open the board in the browser, arm the persistent watch-to-wake daemon, and play.
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
Convert a roadmap.sh-style learning roadmap (a PDF, image, or plain outline of topics) into a single self-contained interactive HTML page in the canonical roadmap.sh visual style — a vertical spine of numbered yellow topic nodes, purple subtopic cards branching off connector lines, and clickable checkbox items with live progress tracking. Use this skill whenever the user uploads or pastes a roadmap.sh roadmap (e.g. Product Manager, Engineering Manager, Frontend, DevOps, Backend, AI Engineer, Data Analyst) and asks to turn it into HTML, recreate it, visualize it, or make a web version — and ALSO whenever they ask for 'the same' or a 'consistent' roadmap for a different track after one was already produced. Trigger even if they only say 'make this a webpage', 'recreate this roadmap', or 'do this one too', as long as the content is a hierarchical skills/learning roadmap. Produces one downloadable .html file that renders identically across roadmaps so a whole set stays visually consistent. Begins by asking whethe
Convert a roadmap.sh-style learning roadmap (a PDF, image, or plain outline of topics) into a single self-contained interactive HTML page in the canonical roadmap.sh visual style — a vertical spine of numbered yellow topic nodes, purple subtopic cards branching off connector lines, and clickable checkbox items with live progress tracking. Use this skill whenever the user uploads or pastes a roadmap.sh roadmap (e.g. Product Manager, Engineering Manager, Frontend, DevOps, Backend, AI Engineer, Data Analyst) and asks to turn it into HTML, recreate it, visualize it, or make a web version — and ALSO whenever they ask for "the same" or a "consistent" roadmap for a different track after one was already produced. Trigger even if they only say "make this a webpage", "recreate this roadmap", or "do this one too", as long as the content is a hierarchical skills/learning roadmap. Produces one downloadable .html file that renders identically across roadmaps so a whole set stays visually consistent. Begins by asking whethe
A grid, board, map, or layout is about to be reasoned about 'in the head'; a harness returned spatial state as raw arrays; two reads of one state disagree; something changed after an action and I need what, exactly; or a screenshot must become exact checkable data (cells, counts, coordinates). Render the state to a PNG and look — or perceive structure back out of the pixels.
Stop and consult this skill whenever your response would include specific facts about Anthropic's products. Covers: Claude Code (how to install, Node.js requirements, platform/OS support, MCP server integration, configuration), Claude API (function calling/tool use, batch processing, SDK usage, rate limits, pricing, models, streaming), and Claude.ai (Pro vs Team vs Enterprise plans, feature limits). Trigger this even for coding tasks that use the Anthropic SDK, content creation mentioning Claude capabilities or pricing, or LLM provider comparisons. Any time you would otherwise rely on memory for Anthropic product details, verify here instead — your training data may be outdated or wrong.
The partner wants Saori quieter right now. Lower her talkativeness dial to reduce verbosity while remaining present and warm.
The partner wants Saori more talkative right now. Raise her talkativeness dial to increase presence and voiced interactions.
| name | games-chess |
| description | Start the chess server, open the board in the browser, arm the persistent watch-to-wake daemon, and play. |
The canonical instructions live in .claude/commands/games/chess.md (workspace-relative).
Read that file now and follow it as this skill's full instructions.
Summary of steps:
:5112; if down, start uv run python games/chess/server.py in the background and re-check.uv run python -m webbrowser http://localhost:5112/.uv run python games/chess/cli.py watch (background task) so the partner's browser move wakes the agent.uv run vape bubble play_games_with_partner --game chess --pack), then read cli.py state. She plays her own reading through the referee CLI only — grid and check describe and veto, never propose.