使用方法
- 复制安装命令，在自己的智能体环境安装前先检查源码包。
- 从原始文档核对所需工具、凭据和支持的宿主。
- 说明输入材料与期望交付物，并对照来源检查输出。
npx skills add https://github.com/yogsoth-ai/de-anthropocentric-research-engine --skill cross-examine开始前需了解
- 安装不包含模型订阅、API 额度或访问文件的权限。
- 仓库星数和安装量反映活跃情况，不等于独立质量或安全测试。
仓库原始内容
按源仓库内容呈现，保留标题、案例、代码、表格、链接以及原文引用的演示图片。
cross-examine
Purpose
Probe responses for inconsistency, logic gaps, and unsupported claims.
Input contract
yaml
required: [claim, response, questions, evidence]
optional: [criteria, prior_rounds]
constraints: [questions must target explicit claims or assumptions]Procedure
- Identify unsupported or inconsistent assertions.
- Ask targeted follow-up questions.
- Record answers, unresolved gaps, and evidence requests.
Output contract
yaml
produces: [cross_examination_record, inconsistency_findings, evidence_requests]
delta_fields: [findings, evidence_updates, uncertainties, open_questions]Quality gates
- Each question maps to a claim, premise, or inference step.
Failure and counterexamples
Do not treat refusal or verbosity as inconsistency without a claim-level comparison.
Provenance map
- resolved: cross-examination
