Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →面向中国商标申请的类别规划、可注册性初筛及申请材料准备技能。基于尼斯分类（国际商标分类），引用中国法律法规。本技能应在接到商标咨询、需要结构化输出建议与风险分级、或需要准备申请材料时使用。不要用于：替代正式法律意见、承诺注册成功率、处理复杂商标争议案件。
.claude/skills/thomasmoreai-trademark-assistant/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | 122% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 99% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 31% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 151% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 139% | 0% |
提供”商标类别规划 + 可注册性初筛”的标准化服务流程，直接产出可交付给客户或内部团队的结构化结论。
主要适用场景： 中国商标申请
法律依据： 本技能引用的法律法规均为中国国内法律法规，包括但不限于《中华人民共和国商标法》《商标法实施条例》及《商标审查审理指南》等。
国际适用性说明：
服务主体： 本技能由中国执业律师提供法律服务支持，仅具备中国大陆法律服务资质。
在以下场景触发本技能：
优先按 references/service-intake-checklist.md 收集信息；缺失关键信息时先提问，不直接下结论。
最少输入：
类别规划 / 可注册性初筛 / 组合服务references/classification-planning-guide.md 完成类别规划references/registrability-prescreen-guide.md 完成可注册性初筛references/output-contract.md 输出标准化结果当客户确认设计方案后，进入申请材料准备阶段：
archive/ 目录references/trademark-examination-and-adjudication-guidelines/chapter-03.md）ywxlaw）references/service-intake-checklist.mdreferences/classification-planning-guide.mdreferences/registrability-prescreen-guide.mdreferences/output-contract.mdreferences/trademark-description-guide.mdreferences/legal-basis-index.mdreferences/trademark-examination-and-adjudication-guidelines/trademark-examination-and-adjudication-guidelines-index.mdreferences/nice-classification-v13-2026/nice-classification-v13-2026-index.mdreferences/nice-classification-v12-2025/nice-classification-v12-2025-index.mdtemplates/导入商品信息.xlsx当需要输出可导入商标系统的商品清单时，使用 templates/导入商品信息.xlsx 模板生成 Excel 文件。
| 列名 | 说明 | 示例 | |------|------|------| | 序号 | 商品序号 | 1, 2, 3... | | 商品类别 | 商标类别编号 | 28 | | 类似群 | 类似群编号 | 2802 | | 商品名称 | 规范商品名称 | 玩具 |
references/nice-classification-v13-2026/class-XX.md 中查找）{商标名}-第{X}类-商品清单.xlsx使用 openpyxl 库生成 Excel 文件：
pythonfrom openpyxl import load_workbook # 加载模板 wb = load_workbook('templates/导入商品信息.xlsx') sheet = wb.active # 填充数据（从第2行开始） for idx, item in enumerate(goods_list, start=2): sheet[f'A{idx}'] = idx - 1 # 序号 sheet[f'B{idx}'] = item['类别'] # 商品类别 sheet[f'C{idx}'] = item['类似群'] # 类似群 sheet[f'D{idx}'] = item['商品名称'] # 商品名称 wb.save('输出文件.xlsx')
商标说明是商标注册申请的必要材料，用于描述商标特征、构成要素及含义。所有商标申请均需撰写商标说明。
对于图形商标或组合商标，必须先使用图像理解工具分析商标图片，再撰写说明。
| 工具 | 类型 | 适用场景 | |------|------|----------| | mcp__zai-mcp-server__analyze_image | MCP | 通用图像分析，推荐优先使用 | | mcp__MiniMax__understand_image | MCP | 备选方案 | | Read 工具直接读取图片 | 内置 | Claude 原生视觉能力 |
使用图像理解工具时，需提取以下信息：
text请详细描述这个商标的构成要素： 1. 商标类型（纯文字/纯图形/组合商标） 2. 文字内容（中文/英文/拼音等） 3. 图形元素（如有） 4. 字体设计特点（普通字体还是艺术字/设计字体，具体设计特征） 5. 颜色使用（如有） 6. 整体设计风格及各元素之间的统一性
撰写指南和示例见 references/trademark-description-guide.md
当完成具体商标申请方案后，将生成的材料归档到 archive/ 目录，便于后续查阅和复用。
archive/
└── {YYYYMMDD}_{商标名}/
├── 申请方案.md # 完整申请方案（含类别规划、风险分析）
├── 商品清单.xlsx # 可导入商标系统的商品清单
├── 商标说明.txt # 商标说明文本
└── notes.md # 备注信息（可选）| 依赖 | 安装方式 | |------|----------| | 无 | 本技能为文本推理与结构化输出流程，无额外系统依赖 |
| 包名 | 用途 | 安装命令 | |------|------|----------| | openpyxl | 生成 Excel 商品清单 | uv run --with openpyxl python script.py |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-21 | pass→pass | 12,395 | 6,308 | -49% | 1 | 1 | 0% | 2,801 | 3,725 | +33% | 0 | 0 | — |
case-19 | fail→pass | 12,034 | 11,946 | -1% | 1 | 1 | 0% | 2,001 | 4,436 | +122% | 0 | 0 | — |
case-20 | fail→pass | 13,333 | 12,020 | -10% | 1 | 1 | 0% | 2,144 | 4,268 | +99% | 0 | 0 | — |
case-01 | fail→fail | 19,062 | 18,475 | -3% | 1 | 1 | 0% | 3,535 | 5,683 | +61% | 0 | 0 | — |
case-02 | fail→fail | 14,533 | 7,747 | -47% | 1 | 1 | 0% | 2,987 | 2,738 | -8% | 0 | 0 | — |
case-03 | fail→fail | 9,494 | 5,671 | -40% | 1 | 1 | 0% | 1,807 | 2,626 | +45% | 0 | 0 | — |
case-04 | fail→fail | 18,313 | 15,635 | -15% | 1 | 1 | 0% | 3,177 | 4,965 | +56% | 0 | 0 | — |
case-05 | fail→pass | 15,634 | 8,119 | -48% | 1 | 1 | 0% | 2,775 | 3,633 | +31% | 0 | 0 | — |
case-06 | fail→fail | 11,140 | 18,478 | +66% | 1 | 1 | 0% | 2,079 | 5,624 | +171% | 0 | 0 | — |
case-07 | pass→pass | 16,452 | 22,787 | +39% | 1 | 1 | 0% | 2,814 | 5,998 | +113% | 0 | 0 | — |
case-08 | fail→fail | 13,197 | 14,462 | +10% | 1 | 1 | 0% | 2,169 | 4,722 | +118% | 0 | 0 | — |
case-09 | fail→pass | 14,252 | 25,571 | +79% | 1 | 1 | 0% | 2,470 | 6,189 | +151% | 0 | 0 | — |
case-10 | fail→pass | 11,925 | 12,895 | +8% | 1 | 1 | 0% | 1,910 | 4,560 | +139% | 0 | 0 | — |
case-11 | fail→pass | 16,140 | 16,422 | +2% | 1 | 1 | 0% | 2,903 | 5,282 | +82% | 0 | 0 | — |
case-12 | fail→pass | 11,843 | 4,138 | -65% | 1 | 1 | 0% | 2,362 | 3,148 | +33% | 0 | 0 | — |
case-13 | pass→pass | 9,582 | 7,690 | -20% | 1 | 1 | 0% | 1,640 | 3,545 | +116% | 0 | 0 | — |
case-14 | fail→pass | 14,439 | 16,955 | +17% | 1 | 1 | 0% | 2,298 | 5,210 | +127% | 0 | 0 | — |
case-15 | fail→pass | 8,335 | 2,986 | -64% | 1 | 1 | 0% | 1,495 | 2,746 | +84% | 0 | 0 | — |
case-16 | fail→pass | 18,148 | 13,050 | -28% | 1 | 1 | 0% | 3,162 | 4,555 | +44% | 0 | 0 | — |
case-17 | fail→fail | 19,833 | 20,888 | +5% | 1 | 1 | 0% | 3,369 | 6,046 | +79% | 0 | 0 | — |
case-18 | fail→pass | 11,256 | 10,721 | -5% | 1 | 1 | 0% | 1,873 | 4,193 | +124% | 0 | 0 | — |
case-22 | pass→pass | 19,611 | 17,081 | -13% | 1 | 1 | 0% | 3,457 | 5,053 | +46% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +50 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.