Thesis Figure Skill
0xE1337/thesis-figure-skill
生成学术论文配图:LaTeX/TikZ 代码(结构化图表,直接嵌入论文)或 draw.io XML (技术路线图、汇报配图)。自动按论文领域风格设计,编译验证后交付。
把 LeetCUDA kernels/interview 教学源码(base/sgemv/sgemm/hgemm/flashattn/ffpaattn/fp8gemm.cuh/fp4gemm.cuh)+ ffpa-attn CuTe sm120 源码(csrc/cuffpa/cute fp8/fp4,commit 861d75e)写成中文 CUDA 技术书(6 Part 38 章 + CuTe…
$ npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install xlite-dev/LeetCUDA write-leetcuda-book --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/xlite-dev/LeetCUDA.git skills-src && mkdir -p .claude/skills && cp -r skills-src/kernels/interview/book/skills/write-leetcuda-book .claude/skills/write-leetcuda-book && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "write-leetcuda-book" agent skill from https://github.com/xlite-dev/LeetCUDA/tree/main/kernels/interview/book/skills/write-leetcuda-book into .claude/skills/write-leetcuda-book/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "write-leetcuda-book", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/xlite-dev/LeetCUDA/tree/main/kernels/interview/book/skills/write-leetcuda-bookType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install xlite-dev/LeetCUDA write-leetcuda-book --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/xlite-dev/LeetCUDA.git skills-src && mkdir -p .agents/skills && cp -r skills-src/kernels/interview/book/skills/write-leetcuda-book .agents/skills/write-leetcuda-book && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "write-leetcuda-book" agent skill from https://github.com/xlite-dev/LeetCUDA/tree/main/kernels/interview/book/skills/write-leetcuda-book into .agents/skills/write-leetcuda-book/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "write-leetcuda-book", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install xlite-dev/LeetCUDA write-leetcuda-book --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/xlite-dev/LeetCUDA.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/kernels/interview/book/skills/write-leetcuda-book .cursor/skills/write-leetcuda-book && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "write-leetcuda-book" agent skill from https://github.com/xlite-dev/LeetCUDA/tree/main/kernels/interview/book/skills/write-leetcuda-book into .cursor/skills/write-leetcuda-book/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "write-leetcuda-book", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/xlite-dev/LeetCUDA.git --path kernels/interview/book/skills/write-leetcuda-book--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install xlite-dev/LeetCUDA write-leetcuda-book --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/xlite-dev/LeetCUDA.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/kernels/interview/book/skills/write-leetcuda-book .gemini/skills/write-leetcuda-book && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "write-leetcuda-book" agent skill from https://github.com/xlite-dev/LeetCUDA/tree/main/kernels/interview/book/skills/write-leetcuda-book into .gemini/skills/write-leetcuda-book/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "write-leetcuda-book", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install xlite-dev/LeetCUDA write-leetcuda-bookInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/xlite-dev/LeetCUDA.git skills-src && mkdir -p .github/skills && cp -r skills-src/kernels/interview/book/skills/write-leetcuda-book .github/skills/write-leetcuda-book && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "write-leetcuda-book" agent skill from https://github.com/xlite-dev/LeetCUDA/tree/main/kernels/interview/book/skills/write-leetcuda-book into .github/skills/write-leetcuda-book/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "write-leetcuda-book", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install xlite-dev/LeetCUDA write-leetcuda-book --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/xlite-dev/LeetCUDA.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/kernels/interview/book/skills/write-leetcuda-book .opencode/skills/write-leetcuda-book && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "write-leetcuda-book" agent skill from https://github.com/xlite-dev/LeetCUDA/tree/main/kernels/interview/book/skills/write-leetcuda-book into .opencode/skills/write-leetcuda-book/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "write-leetcuda-book", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
write-leetcuda-book把 LeetCUDA kernels/interview 教学源码(base/sgemv/sgemm/hgemm/flashattn/ffpaattn/fp8gemm.cuh/fp4gemm.cuh)+ ffpa-attn CuTe sm120 源码(csrc/cuffpa/cute fp8/fp4,commit 861d75e)写成中文 CUDA 技术书(6 Part 38 章 + CuTe…
Write Leetcuda Book is an agent skill from xlite-dev/LeetCUDA. 把 LeetCUDA kernels/interview 教学源码(base/sgemv/sgemm/hgemm/flashattn/ffpaattn/fp8gemm.cuh/fp4gemm.cuh)+ ffpa-attn CuTe sm120 源码(csrc/cuffpa/cute fp8/fp4,commit 861d75e)写成中文 CUDA 技术书(6 Part 38 章 + CuTe 白皮书导读 + 5 附录,XeLaTeX→PDF,终态 585 页零 error)。RFC-0..N 已全部完成(2026-09-24,Part V=FP8/FP4 HGEMM 篇 ch34-37 新增,原 FP8/FP4 Attention 篇顺延为第六部分),当前任务形态 = 增补章(RFC-L/M/N 模式)/复审/勘误/图重绘。当任务涉及:撰写或修改书稿章节(chapters/chNN.tex)、每章最小测试(book/tests/chNN.cu)、ffpaattn.bench 验证、知乎资料收集与提炼、TikZ inline 图新建(主路径)与存量 drawio 图维护、源码注释核查、勾选 RFC 进度、构建/审校…
Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. The skill folder holds 8 other files (for example `BOOK_PLAN.md`, `RFC.md` and `examples/drawio-recon-m1/README.md`).
It sits in Documents & Office, covering Diagrams, LaTeX and GPU and accelerator computing. It works with CUDA, draw.io, LaTeX and C++. The repository describes itself as: An Open-source book with Modern CUDA learning Notes for Beginners - BF16/FP8/FP4, HGEMM, CuTe, Flash-Attention, etc. The licence is GPL-3.0.
5 steps, taken from the first numbered list in SKILL.md.
Read from SKILL.md and the folder at commit c5e985f. It shows what the files ask for, not the result of running them.
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Shell commands in SKILL.md call:
python3pdftoppmFrom the folder's file list and the shell code blocks in SKILL.md.
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Write Leetcuda Book loads about 2k tokens when it runs. Until then it costs about 164 tokens; SKILL.md has 464 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
The full file from xlite-dev/LeetCUDA at commit c5e985f, republished under its GPL-3.0 licence (© xlite-dev). 464 words, ~2,001 tokens.
.claude/skills/write-leetcuda-book/SKILL.md (or your agent's skills folder). This skill also uses 6 other files; get the full folder from GitHub.本 skill 目录即唯一事实源,两份核心文档(同目录):
| 文件 | 角色 |
|---|---|
BOOK_PLAN.md | 规范源:5 Part 35 章+导读结构、每章模板与 DoD 八条(含 Part V 特化 DoD)、素材/测试/图表/公式工程、LaTeX 管线、风险与验收 |
RFC.md | 执行跟踪:RFC-0..M 里程碑 checkbox(B 已取消)、每章执行卡片、图清单登记表、知乎参考种子表 |
绝对路径:/workspace/dev/vipshop/LeetCUDA/kernels/interview/book/skills/write-leetcuda-book/(已软链到 /workspace/dev/vipshop/.github/skills/write-leetcuda-book)。
RFC.md 里程碑总览表 + 对应章节的执行卡片。RFC-0..M 已全部收口(RFC-B 于 2026-09-18 取消,职能归附录 E + drawio 全量重建),存量只剩零星复审项(如 B.5 RoPE 参考);新需求 = 用户新点名的增补章/增强项,按 RFC-L/M 模式新开里程碑条目(源码整合→测试→bench→正文→接线→验收→code review→用户增强)逐项推进。BOOK_PLAN.md §3 章节卡片(源码区间/宏/公式/图/参考/测试映射)+ §4 模板与 DoD + 相关工程节。remove-tech-ai-tone skill 做文风自检(八类 AI 味模式 + 词表逐词 grep,存疑保留;图内文字只做等宽替换),通过后才进入下一步。- [x] + 追加日期;发现问题记 CHECKLOG / RFC 对应位置。.cuh + bench/ 下 10 个 .cu(bench/bench_leetcuda.cu 与 9 个模块 .cu:bench/{base,sgemv,sgemm,hgemm,fp8_gemm,fp4_gemm,flash_attn,ffpa_attn,utils}.cu,2026-09-29 迁入 bench/ 后均已登记)——在 RFC-A 登记 SHA256 后零改动。注释错误只记 book/CHECKLOG.md + 正文「勘误与考据」,不回写源码。linerange 多段裁剪,firstnumber=auto);完整代码给附录 D 的 GitHub commit permalink;每章正文(非 listings)≥3500 字。book/tests/chNN_*.cu 无 cuBLAS/cuDNN 依赖,CPU fp64 参考 + 容差三档(F32Acc 1e-3 / F16Acc 5e-2 / TF32 1e-2),规模 ≤512,arch 不在位输出 SKIP;行数 ≤300(基础章)/≤500(复杂章)。zhihu-analysis/→提炼→成文不照抄→附录 E 汇总);图片直接引用+出处标注(作者/文章/链接/日期),原图归档 book/figures/zhihu/;水印/作者角标/平台 logo 一律标记为非内容元素,不得进入书内图。tikz-diagrams skill(模板/编译渲染/视觉 QA 工具链;ch26b 六图先例,上游 5be940f 亦将存量图大规模迁为 tikzpicture)。drawio 降为存量维护路径(ch20-32 仍 36 处 png 引用):drawio-reconstruction skill 为主(inventory→重建→审查闭环),drawio-diagram-builder/drawio-flow-forge 辅助。执行模式:主 agent 直接做,不派 task agent(闭环子代理实测易卡死,用户 2026-09-11 拍板)。审查结论如实记 audit(自审需标注)。drawio 质量三查(XML well-formed + COVER=0 + 渲染宽不越界,见「drawio 制图铁律」);.drawio 源文件必须入库(用户要求可手改,勿只提交 PNG)。
⚠️ TikZ 图强制视觉验收(2026-09-28 用户定为铁律):TikZ 图出现文文重叠/图文重叠极其常见(坐标估算的文字宽度与真实渲染字宽不符所致,N.19 七图 + O.18 六图全是此类)。任何新建/修改 tikzpicture 后,必须真实渲染并 view_image 逐角落检查确认无重叠后才算完成——只看代码/编译零 error 不算验收。配方:抽图到 standalone(border=8pt + \usepackage[UTF8,fontset=fandol]{ctex},fandol 失败则 fontset=none + \setCJKmainfont{Noto Sans CJK SC},CJK 文本必须真实渲染、禁止用假文本代替——重叠取决于真实字宽)+ 从 book/preamble.tex 复制用到的颜色/样式定义 → xelatex -interaction=nonstopmode → pdftoppm -r 150 -png → view_image;修复优先级:平移节点/换 anchor/text width+align 换行/箭头改走线/竖排改横排(横排更宽,要挪到空白带)/字号降档(≥\scriptsize)。修法只动排版不动内容。foreach 标签里的循环变量(如 16\i)会被文本替换成数字,语义上想保留符号 i 必须改用显式值表({0/0/16,1/16/32,...})。book/,TEXMFCNF=../tex/: xelatex -interaction=nonstopmode book.tex ×2(或 ./build.sh);本机已有 xelatex+ctexbook+Noto CJK(已验证)。TOC 全量装载报 main_memory capacity exceeded(PDF 目录丢失)时须 fmtutil-sys --byfmt xelatex 重建 fmt——texmf.cnf 的 main_memory 仅 fmt 生成期生效,改配置不重建无效(2026-09-22 L.9 根因)。# drawio headless 导出(已装 v31.4.5 deb + xvfb;Ubuntu noble 无 apt 包,须用 GitHub releases deb)
DRAWIO_PATH=/usr/local/bin/drawio-headless \
python3 /workspace/dev/vipshop/.github/skills/drawio-reconstruction/scripts/export_drawio.py in.drawio out.png
python3 /workspace/dev/vipshop/.github/skills/drawio-reconstruction/scripts/check_drawio.py in.drawio
# 也可直接:/usr/local/bin/drawio-headless -x -s 3 -o out.png in.drawio
# VS Code drawio 插件(用户已装):用于人工查看/微调 .drawio;自动导出仍走上面 CLI
# 全书图几何体检(book/scripts/,2026-09 新增;改动后必须复检至 0)
python3 scripts/check_overlap.py figures/drawio/*/*.drawio # 文字覆盖(COVER = 后画实心框盖住先画 text)
python3 scripts/check_text_fit.py figures/drawio/*/*.drawio # 渲染宽 vs geometry 宽 + 页宽越界
# 知乎 CLI(已授权;--count 上限 10)
/root/.local/share/zhihu-cli/current/zhihu-cli search zhihu --query "..." --count 10
python3 /workspace/dev/vipshop/.github/skills/zhihu/scripts/fetch_column_fulltext.py <专栏URL> # 枚举/取全文
# 源码锚点断言(RFC-A 后可用)
python3 /workspace/dev/vipshop/LeetCUDA/kernels/interview/book/scripts/verify_anchors.py --ch chNN生成器结构:每图一份自包含 gen.py(F 类 box/text/line + 输出 python3 gen.py <输出文件全名>,如 gen.py fig-12-2.drawio——误传目录名会产生孤儿文件)。批量期的权威源在 .tmp/drawio/gen_batch_{a,b1,b2,c1,c2,d,e}.py(.gitignore 忽略),每图目录的 gen.py 是其回流拷贝(入库、自包含)。改图必须改 .tmp 权威源再回流,直接改目录版会在下一轮批量重跑时被静默覆盖(fig-14-2 实测踩中);回流后用 cmp -s 校验。
XML 铁律:
value 禁裸 < / &:Tile<64,64,16>、Swizzle<3,4,3>、->、R^16 类写法的 < 必须转义 <。裸 < 毁掉 well-formed → drawio 静默渲染成细条(fig-26-1 只输出 2522x74,书里等于废图)。ET.fromstring(xml) 自检;全书定期 python3 -c "ET.parse" 全量扫(本次 2/39 破损)。text 元素不自动换行:geometry 宽小于渲染宽,文字直接溢出色框(甚至越出页宽)。长注释必须手工拆多行,或把 geometry 宽改到 ≥ 估算渲染宽。fontSize×1.02,ASCII × fontSize×0.58(含全角标点按 CJK 计)。check_text_fit.py 用此公式报 [OOB](越页宽)/[COLLIDE](侵入右邻框)。check_overlap.py 的 COVER 即此类)。PNG 宽 ≈ pageWidth × 3(-s 3),高宽比 ≈ pageHeight/pageWidth。比例异常 = 渲染崩了(先查 XML well-formed),不要直接 includegraphics。常见布局病(本轮回修):说明行(y 40–60)与列头框(y 50+)净空不足 → 原点下移 16–26px;卡片内文字 409px 渲染塞 300px 框 → 框加宽 + 拆行;左表右侧的 ... 居左文本 geometry 过宽侵入相邻框 → 宽度收到渲染宽;「标题 + 图」跨页 → 图示节前 \needspace{图高+30mm}(mm = width_tw × 181 × png_h/png_w,preamble.tex 已 \usepackage{needspace})。
其它约定:CJK 禁 bold(易糊);两位数格宽 ≥54px;纯色缩略格无文字;.drawio/.png/gen.py 三件套全部入库。
../../)kernels/interview/book/
├── book.tex / preamble.tex / build.sh # RFC-0 产出
├── chapters/ch00-ch33 + ch19b/ch26b/ch26c 增补 + wp/ 白皮书导读 … appendices/ # RFC-C..M 产出
├── figures/{ascii,zhihu,drawio,ffpa,misc,tikz}/
├── tests/{common_test.h, chNN_*.cu, build_tests.sh}
├── notes/ references/ scripts/ CHECKLOG.md
├── skills/write-leetcuda-book/ # 本 skill(规范+跟踪)
└── .tmp/ # 测试/bench 临时产物(按任务建子目录)examples/drawio-recon-m1/:drawio 重建 smoke test 固化样例(重建件 + headless 导出预览 + audit + 复现命令)。正式重建任务照此流程执行;原图在 drawio-reconstruction skill 的 examples/m1.png(路径引用,不复制)。remove-tech-ai-tone(文风自检(强制):新章/增补章完成、commit 前必跑——技术文去 AI 味八类模式 + 高频词表 + 保留词表 + 验证工作流;目标文风「精确、简洁、沉稳、直接」。全书 2026-09-28 四轮扫描(~305 处)沉淀的词表源于此 skill)tikz-diagrams(图主路径:TikZ inline 写在章节 tex 内——全书 36 章已含 176 处 tikzpicture(ch26b/ch26c 全 TikZ 先例 + 上游 5be940f 大迁移);提供模板/编译渲染/视觉 QA 工具链,产出 .tex+.pdf+.png)drawio-reconstruction(drawio 存量图维护:ch20-32 仍有 36 处 drawio png 引用,重建走 inventory→审查闭环)、drawio-diagram-builder / drawio-flow-forge(存量 drawio 修改/新建)zhihu(资料五步法)cuda-cpp-kernel(ptx-docs=F2 核查源)、cutlass-cpp-kernel(arch guides=F1 核查源)、cuda-auto-tune(bench 复测方法论)ffpa-cuda-understand(ch19/26/26c 大 D split-D 背景)© xlite-dev, GPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
SKILL.md and 6 other files in kernels/interview/book/skills/write-leetcuda-book of xlite-dev/LeetCUDA.
Open the folder on GitHubat commit c5e985f
Write Leetcuda Book next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Write Leetcuda Book this skillxlite-dev/LeetCUDA | 12k | — | ~2k | Automated safety check: Pass | GPL-3.0 | |
| Thesis Figure Skill0xE1337/thesis-figure-skill | 144 | — | ~9.1k | Automated safety check: Pass | MIT | |
| Hf Architecture Tikzyzlnew/infra-skills | 149 | — | ~2k | Automated safety check: Pass | None | |
| Nsfc RoadmapInternScience/DrClaw | 172 | — | ~2.5k | Automated safety check: Notes | None | |
| Nsfc SchematicInternScience/DrClaw | 172 | — | ~3.7k | Automated safety check: Notes | None | |
| Legal Visualizationcat-xierluo/legal-skills | 720 | — | ~2k | Automated safety check: Warn | CC-BY-NC-4.0 |
0xE1337/thesis-figure-skill
生成学术论文配图:LaTeX/TikZ 代码(结构化图表,直接嵌入论文)或 draw.io XML (技术路线图、汇报配图)。自动按论文领域风格设计,编译验证后交付。
yzlnew/infra-skills
Draw Sebastian-Raschka-gallery-style TikZ architecture diagrams for any HuggingFace decoder-only LLM, with per-block parameter formulas and concrete numbers.
InternScience/DrClaw
当用户明确要求"生成 NSFC 技术路线图/技术路线图绘制/roadmap/flowchart"或需要把标书研究内容转成"可打印、A4 可读"的技术路线图时使用。默认输出可编辑源文件(.drawio)与可嵌入文档的渲染结果(.svg/.png/.pdf);当用户主动提及 Nano Banana/Gemini 图片模型时,可切换为 PNG-only 模式。⚠️…
InternScience/DrClaw
当用户明确要求"生成 NSFC 原理图/机制图/schematic diagram/mechanism diagram"或需要把标书中的研究机制、算法架构、模块关系转成"可编辑 + 可嵌入文档"的图示时使用。默认输出可编辑源文件(.drawio)与渲染文件(.pdf/.svg/.png);当用户主动提及 Nano Banana/Gemini 图片模型时,可切换为 PNG-only 模式。⚠️…
cat-xierluo/legal-skills
Legal Visualization。面向法律业务场景的法律图解与图表生成技能;当用户要求把案件材料、合同材料、合规事项、交易安排、证据链、诉讼流程、时间轴、法律关系、客户汇报、法律服务方案或律师团队工作整理成关系图、流程图、时间轴、证据链、风险图、路线图、PNG/SVG/PDF/.drawio…
wentorai/research-plugins
Create publication-quality scientific diagrams with TikZ in LaTeX
xlite-dev/LeetCUDA
LeetCUDA 中文技术书(584 页,XeLaTeX 源)按需查阅 skill——写、优化、调试或 review CUDA C++/PTX kernel 时的权威参考路由层。当任务涉及:GPU 架构/Roofline/ occupancy、向量化与 coalescing、warp/block reduce、softmax(online/LSE merge)、…
Categories
把 LeetCUDA kernels/interview 教学源码(base/sgemv/sgemm/hgemm/flashattn/ffpaattn/fp8gemm.cuh/fp4gemm.cuh)+ ffpa-attn CuTe sm120 源码(csrc/cuffpa/cute fp8/fp4,commit 861d75e)写成中文 CUDA 技术书(6 Part 38 章 + CuTe…. Write Leetcuda Book is an agent skill from xlite-dev/LeetCUDA.
Write Leetcuda Book fits situations like: tasks that involve Diagrams; tasks that involve LaTeX; tasks that involve GPU and accelerator computing.
Run `npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a claude-code`. Or copy the skill folder (kernels/interview/book/skills/write-leetcuda-book in xlite-dev/LeetCUDA) into .claude/skills/write-leetcuda-book in your project. Claude Code loads it when a task matches its description.
Run `npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a codex`. Or copy the skill folder (kernels/interview/book/skills/write-leetcuda-book in xlite-dev/LeetCUDA) into .agents/skills/write-leetcuda-book in your project. Codex loads it when a task matches its description.
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add xlite-dev/LeetCUDA --skill write-leetcuda-book -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/write-leetcuda-book, .gemini/skills/write-leetcuda-book, .github/skills/write-leetcuda-book and .opencode/skills/write-leetcuda-book in your project.
Going by SKILL.md and its folder, Write Leetcuda Book needs the command-line tools its instructions call (python3 and pdftoppm). Our summary lists: Python 3.
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
Write Leetcuda Book is published under the GPL-3.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
About 2k tokens (SKILL.md is roughly 8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
Skills that share tags, products or a category with Write Leetcuda Book: Thesis Figure Skill (0xE1337/thesis-figure-skill, 144 stars), Hf Architecture Tikz (yzlnew/infra-skills, 149 stars), Nsfc Roadmap (InternScience/DrClaw, 172 stars) and Nsfc Schematic (InternScience/DrClaw, 172 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
xlite-dev (a GitHub organization) maintains it in xlite-dev/LeetCUDA, which has 12,035 GitHub stars. The repository holds 2 skills in this directory. The repository was last updated on September 30, 2026.
Source: xlite-dev/LeetCUDA on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.