凡自治之物 · LGD 治理学 · 创造术家族 ｜ 评测协议：skill-quality-gate
English
When to use
- Before an AI system or agent goes live, when you need to quickly produce a standardized document stating its identity and intended use.
- When internal compliance review requires a transparency report, data-usage scope and model-limitations information.
- When team collaboration requires storing model metadata in both Markdown and JSON formats for archival.
Usage
- Run the core script
model_card_generator.py; it must be executed inside a Python environment where the project directory contains the script. - Pass the required input parameters, such as model name, data-source description, and the risk points to highlight.
- When the skill is consumed as an interface, pass the task configuration and context information via JSON-RPC.
Workflow
- Collect the model's training objective, main use cases and key functional identity-definition information.
- Organize risk elements such as data-privacy scope, known biases and system performance limits.
- Run the generation script to output a model-card draft that meets transparency obligations.
- Check that the generated Markdown and JSON include contact information and the disclaimer section.
- Save the final document to the project repository or deployment-package directory for team review.
Evidence & safety statement
- Evidence: TBD (no evaluation / badge link yet; update here once produced)
- Version: 1.0.1
- Safety: no independent security scan has been performed (TBD); the conclusion is pending and no endorsement is given until it is completed.
中文
何时用
- 在 AI 系统或智能体上线前，需要快速生成包含身份与用途说明的标准化文档时。
- 内部合规审查要求提供透明度报告、数据使用范围及模型限制信息时。
- 团队协作中需以 Markdown 和 JSON 双格式存储模型元数据并归档时。
用法
- 执行命令调用核心脚本
model_card_generator.py，需处于 Python 环境且项目目录下存在该文件。 - 传入必要的输入参数，如模型名称、数据源描述及需要强调的风险点。
- 若技能作为接口使用，则通过 JSON-RPC 方式传递任务配置与上下文信息。
流程
- 收集模型的训练目标、主要用途及关键功能身份定义信息。
- 整理数据隐私范围、已知偏差及系统性能限制等风险要素。
- 运行生成脚本，输出符合透明度义务要求的模型卡草稿。
- 检查生成的 Markdown 与 JSON 内容是否包含联系人信息与声明章节。
- 将最终文档保存至项目存储库或部署包目录中供团队查阅使用。
证据与安全声明
- 证据：TBD（暂无评测/徽章链接；产生后在此更新实际链接）
- 版本：1.0.1
- 安全：未做独立安全扫描（TBD），结论待补；补全前不作推荐背书。