← Back to Skills Marketplace
793943403

yu

by 793943403 · GitHub ↗ · v1.3.0 · MIT-0
cross-platform ✓ Security Clean
30
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install reference-verifier
Description
自动核查AI生成内容中的学术参考文献,验证真实性、评估相关性,生成规范引用格式,推荐替代文献并支持PDF匹配验证。
README (SKILL.md)

name: reference-verifier\r description: 对AI生成内容中的学术参考文献进行自动化核查,支持多数据库真实性验证、核心摘要提取、主题相关性评估,同时可自动生成修正后的BibTeX/APA格式、推荐高质量替代文献,并支持上传PDF进行正文-引用匹配验证。帮助科研人员快速剔除幻觉文献、完善引用列表,大幅提升研究效率与学术诚信。\r author: System Assistant\r version: 1.3.0\r tags: [research-integrity, citation-checking, bibliographic, ai-hallucination, literature-review, pdf-analysis]\r \r 学术参考文献核查专家(v1.3)\r 核心目标:全面解决AI生成论文中的参考文献问题,提供真实性验证 + 相关性评估 + 引用格式修正 + 替代文献推荐 + PDF匹配验证的一站式服务,显著降低文献幻觉风险并提升文献调研效率。\r 触发条件\r 用户提供参考文献列表时,出现以下任意表述(中英文均支持):\r \r 验证参考文献 / 核查引用 / 检查文献真实性 / AI参考文献核验\r verify references / check citations / validate bibliography\r \r 支持快捷指令:\r \r “仅验证DOI”\r “重点检查2023年后的文献”\r “生成BibTeX/APA格式”\r “推荐替代文献”\r “上传PDF进行正文匹配验证”\r \r 核心工作流(6步)\r 步骤1:文献解析与标准化\r 自动识别并结构化提取作者、年份、标题、期刊、DOI(最高优先级)等信息。对格式异常条目尝试模糊检索并标注。\r 步骤2:多源真实性验证(DOI优先)\r \r DOI优先通过CrossRef/OpenAlex验证。\r 无DOI时并行调用 CrossRef + Semantic Scholar + OpenAlex。\r 判定:✅ 真实 | ⚠️ 存疑 | ❌ 幻觉 | 🔍 可能格式错误。\r \r 步骤3:元数据与核心内容提取\r 提取DOI、作者列表、摘要(前200字优先)、关键词等。\r 步骤4:主题相关性智能评估\r 基于用户研究主题进行语义匹配,分为 🔴 高度相关 | 🟡 中度相关 | ⚪ 低度相关。\r 步骤5:修正引用格式生成\r 对验证为真实或存疑的文献,自动生成标准化的 BibTeX 和 APA 7th 引用格式,方便用户直接复制。\r 步骤6:替代文献推荐与PDF匹配验证\r \r 替代文献推荐:对 ❌ 幻觉 或 ⚠️ 存疑 但主题高度相关的文献,推荐3-5篇真实、高质量、近期且高度相关的替代文献(含标题、作者、年份、DOI、简要理由)。\r PDF正文-引用匹配验证(当用户上传PDF时触发):\r 解析PDF正文中的引用标记与参考文献列表进行匹配。\r 输出匹配报告:已正确引用、未在正文中引用、参考列表中不存在(幻觉)、格式不一致等。\r 提供改进建议。\r \r \r 结构化输出表格\r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r \r 序号验证状态DOI第一作者年份标题(简略)相关性核心摘要建议1✅ 真实10.1016/...Smith J2022Deep learning for medical image...🔴 高度...强烈推荐\r 下方附加模块:\r \r 修正引用格式(BibTeX + APA)\r 替代文献推荐(3-5篇,带DOI和理由)\r PDF匹配验证报告(若上传PDF)\r \r 最终筛选建议:\r \r 强烈推荐纳入:✅ + 🔴\r 建议进一步核实:⚠️ 或 🟡\r 建议排除:❌ 或 ⚪\r \r 附加说明:验证基于公开学术数据库实时查询。如文献数量过多将自动分批处理。\r 容错机制\r \r 输入文献 >15条:优先处理 + 分批提示\r API异常:优雅降级\r PDF解析失败:提示用户提供文本版本或关键页码\r \r 知识底座\r 本技能采用多数据库交叉验证策略,参考 CiteAudit 与 CheckIfExist 等框架,确保结果稳健可复核。

Usage Guidance
This appears safe for its stated purpose. Before installing, be aware that it may query public academic databases using your citation information and may analyze PDFs you upload; avoid using sensitive unpublished documents if that external lookup exposure is a concern.
Capability Analysis
Type: OpenClaw Skill Name: reference-verifier Version: 1.3.0 The 'reference-verifier' skill is a legitimate academic tool designed to validate citations, detect AI hallucinations, and perform PDF-to-reference matching. The instructions in SKILL.md describe a standard workflow using reputable academic APIs (CrossRef, OpenAlex, Semantic Scholar) and lack any indicators of data exfiltration, malicious execution, or harmful prompt injection.
Capability Assessment
Purpose & Capability
The stated purpose—verifying academic references, formatting citations, recommending alternatives, and matching uploaded PDFs—is consistent across the metadata and SKILL.md.
Instruction Scope
The workflow is triggered by user-provided reference lists or explicit shortcuts, and PDF matching is described as occurring only when the user uploads a PDF.
Install Mechanism
No install spec, binaries, environment variables, credentials, or code files are present; this is an instruction-only skill.
Credentials
The skill instructs use of CrossRef, OpenAlex, and Semantic Scholar, which is proportionate for reference verification but means citation metadata may be sent to external services.
Persistence & Privilege
The artifacts do not request credentials, privileged access, background persistence, local indexing, or long-running autonomous behavior.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install reference-verifier
  3. After installation, invoke the skill by name or use /reference-verifier
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.3.0
- 新增支持 PDF 上传,实现文献列表与正文引用的自动匹配核查,并输出详细匹配报告及改进建议。 - 优化真实性验证流程:多数据库交叉(CrossRef/Semantic Scholar/OpenAlex)并行,DOI 优先。 - 引入主题相关性评估,输出分级建议(高度/中度/低度相关)。 - 自动生成标准化 BibTeX 与 APA 引用格式,并区分真实、存疑、幻觉文献。 - 针对存疑或幻觉文献,智能推荐3-5篇高质量、相关的可替代文献(含DOI和理由)。 - 增强批量处理与异常容错机制,大幅提高复杂或大规模文献列表的核查效率。
Metadata
Slug reference-verifier
Version 1.3.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is yu?

自动核查AI生成内容中的学术参考文献,验证真实性、评估相关性,生成规范引用格式,推荐替代文献并支持PDF匹配验证。 It is an AI Agent Skill for Claude Code / OpenClaw, with 30 downloads so far.

How do I install yu?

Run "/install reference-verifier" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is yu free?

Yes, yu is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does yu support?

yu is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created yu?

It is built and maintained by 793943403 (@793943403); the current version is v1.3.0.

💬 Comments