← 返回 Skills 市场

label audit食品标签审核

Name: label audit食品标签审核
Author: cloudyxuq

作者 Tinker · GitHub ↗ · v1.0.1 · MIT-0

cross-platform ✓ 安全检测通过

155

总下载

当前安装

版本数

在 OpenClaw 中安装

/install label-compliance-audit

功能描述

智能审核食品包装标签的合规性，支持图片/PDF设计稿；当用户需要审核标签强制性标识、营养标签规范性、声称用语合规性或格式要求时使用

使用说明 (SKILL.md)

标签合规审核

任务目标

本 Skill 用于：食品包装标签的智能合规审核
能力包含：识别标签内容、对照GB7718/GB28050标准检查、生成审核报告
触发条件：用户上传包装设计稿并要求审核标签合规性

前置准备

依赖说明：无需外部依赖，使用智能体原生能力
非标准文件/文件夹准备：无需额外准备

操作步骤

标准流程：
1. 读取设计稿
  - 使用图像识别工具提取标签中的所有文字和图片内容
  - 识别关键信息：品名、配料表、净含量、生产者信息、营养标签、声称用语等
2. 选择审核模式
  - 模式A：智能体自主审核
    - 基于以下参考标准进行全面审核
    - 生成结构化审核报告
  - 模式B：逐项清单比对
    - 参考 assets/audit-checklist.md 中的审核清单
    - 逐项对照并记录审核结果
3. 执行审核（参考以下文档）
  - 强制性标识完整性：见 references/gb7718-standard.md
  - 营养标签规范性：见 references/gb28050-standard.md
  - 声称用语合规性：见 references/claims-guideline.md
  - 格式要求（字体大小、对比度）：见 references/gb7718-standard.md
  - 过敏原标识：见 references/gb7718-standard.md
4. 生成审核报告
  - 列出所有不符合项
  - 提供整改建议
  - 标注风险等级（严重/一般/轻微）
可选分支：
- 当设计稿为PDF时：先提取页面内容，再进行图像识别
- 当需要详细NRV计算：参考GB28050标准中的计算公式逐项验证

资源索引

领域参考：
- references/gb7718-standard.md（何时读取：审核强制性标识、格式要求、过敏原）
- references/gb28050-standard.md（何时读取：审核营养标签、NRV计算）
- references/claims-guideline.md（何时读取：审核声称用语）
资产模板：
- assets/audit-checklist.md（用途：逐项清单比对审核）

注意事项

审核时需全面识别标签中的所有文字信息，包括小字说明
营养标签的NRV%需根据GB28050中的计算方法逐项验证
声称用语必须满足特定条件才能使用（如"无糖"需≤0.5g/100g）
格式要求需关注字体高度（≥1.8mm）和背景对比度
过敏原标识需明确标注，采用加粗或特殊字体

使用示例

示例1：智能体自主审核

功能说明：上传设计稿，智能体自动完成全面审核
执行方式：智能体读取参考标准，自主判断并生成报告
关键要点：
- 全面识别标签内容
- 逐项对照标准要求
- 生成结构化审核报告
示例：用户上传饼干包装设计稿，要求审核标签合规性

示例2：逐项清单比对

功能说明：使用审核清单逐项核对标签内容
执行方式：参考 assets/audit-checklist.md 逐项检查
关键要点：
- 按清单顺序逐项确认
- 记录每一项的审核结果
- 汇总不符合项
示例：用户要求使用清单方式审核饮料标签

示例3：营养标签专项审核

功能说明：重点审核营养标签的规范性和NRV计算
执行方式：重点读取 references/gb28050-standard.md
关键要点：
- 验证能量、蛋白质、脂肪等核心营养素标注
- 计算并核实NRV%
- 检查营养声称的合规性
示例：用户需要验证酸奶营养标签的NRV%计算准确性

安全使用建议

This skill is internally consistent and contains helpful reference documents and a checklist. Before installing or using it, verify where image/PDF processing (OCR/vision) runs: if the agent will call a third‑party OCR service, avoid uploading sensitive images or confirm the service's privacy policy. Test with non-confidential samples to confirm output quality and check that generated NRV calculations and claims checks match your local/regional regulations. If you need guarantees about where data is processed, ask the vendor or platform how OCR is implemented (local vs external).

功能分析

Type: OpenClaw Skill Name: label-compliance-audit Version: 1.0.1 The skill bundle is a legitimate tool designed for auditing food packaging labels against Chinese national standards (GB7718 and GB28050). The instructions in SKILL.md and the reference documents in the references/ directory are strictly focused on label compliance, OCR text extraction, and report generation, with no evidence of malicious intent, data exfiltration, or harmful prompt injection.

能力评估

✓ Purpose & Capability

Name/description match the included reference documents and checklist (GB7718, GB28050, claims guideline). The skill does not request unrelated credentials, binaries, or config paths — all requested capabilities are appropriate for a label-audit skill.

ℹ Instruction Scope

SKILL.md stays within the stated purpose (extract text/images, compare with standards, produce a report). It references using an image-recognition tool but does not specify which tool or whether processing occurs locally or via a third party; that is a usability/privacy note rather than an incoherence.

✓ Install Mechanism

Instruction-only skill with no install spec and no code files—nothing is written to disk or downloaded during install, which is proportionate for a policy/checklist-based auditor.

✓ Credentials

No environment variables, credentials, or config paths are required. The skill's resource usage (reading included reference docs and user-uploaded designs) is proportionate to its purpose.

✓ Persistence & Privilege

always is false and the skill does not request persistent system-wide privileges or modify other skills. Autonomous invocation is allowed by platform default but is not combined with any elevated access here.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install label-compliance-audit
安装完成后，直接呼叫该 Skill 的名称或使用 /label-compliance-audit 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.1

Version 1.0.1 of label-compliance-audit - No file changes detected in this release. - Documentation and core workflow remain unchanged. - Maintains all previous features for intelligent food label compliance auditing.

v1.0.0

label-compliance-audit v1.0.0 - Initial release: Smart compliance auditing for food packaging labels, supporting image/PDF files. - Checks mandatory elements, nutrition label accuracy, claims wording, format, and allergen identification based on GB7718/GB28050 standards. - Offers two audit modes: automatic full review or checklist-based item-by-item comparison. - Generates structured audit reports highlighting nonconformities, recommendations, and risk levels. - No external dependencies required.

元数据

Slug label-compliance-audit

版本 1.0.1

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 2

常见问题