Fact Check X Complete

面向真实业务决策的多平台事实核验 Skill。正式支持深知晓、深知晓（深度溯源）、豆包、腾讯元宝、DeepSeek 和通义千问，按用户输入动态选择 N≥1 个平台。完整采集原回答、引用、截图与页面存证，结构化比较关键事实，按需使用可信搜索逐点核验权威证据，并评估各平台表现。权威答案是核验产物之一，原始证据、冲突与证据边界全程可追溯。基础采集与对比无需 API Key；可信搜索是权威核验阶段的可选增强。语义工作由 Codex、Claude Code、WorkBuddy 等当前运行载体完成，不调用外部大模型 API。

4 starsSoftware Development

Install

skills CLI (Claude Code, Cursor, Codex and more)

npx skills add ASI2030/Fact-Check-X --skill fact-check-x-complete

Manual install

git clone --depth 1 https://github.com/ASI2030/Fact-Check-X.git /tmp/Fact-Check-X
mkdir -p ~/.claude/skills
cp -r /tmp/Fact-Check-X/skills/fact-check-x-complete ~/.claude/skills/fact-check-x-complete

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# 全知晓（Fact-Check-X）

![Fact-Check-X 多平台事实核验：完整采集、知识点对比、权威核验与答案生成、平台表现评估](https://raw.githubusercontent.com/ASI2030/Fact-Check-X/main/assets/fact-check-x-overview.png?v=1.1.21)

把同一个问题交给一个或多个 AI 平台，完整保留每家的回答和引用，再把关键事实逐点对齐、核验并评估各平台表现。第三步会基于权威证据生成最终答案，但不会用“答案生成”代替完整事实核验：证据冲突会被保留，官方无法查证的内容统一标为“疑似误导”，移入补充参考风险区且不写入确定结论。用户只需说出问题和要比较的平台，不需要学习平台 ID、内部流程编号或报告术语。

## 支持平台

当前内置以下网页端平台，按用户本次输入动态选择任意 `N≥1` 个：

| 平台 | 能力 |
|---|---|
| 深知晓 | 标准问答、引用与官方来源采集 |
| 深知晓（深度溯源） | 等待普通回答完成后在同一会话启动深度溯源，并作为独立结果采集 |
| 豆包 | 回答、引用与页面存证 |
| DeepSeek | 回答、引用、逐条来源正文与页面存证 |
| 通义千问 | 回答、引用、逐条来源正文与页面存证 |
| 腾讯元宝 | 回答、引用、逐条来源正文与页面存证 |
平台由用户选择，不存在固定“五平台模式”。`N=1` 可完成单平台事实核验；`N≥2` 会额外比较各平台的共识、冲突和来源差异。网页结构变化时，技能会明确报告失败并停在当前阶段，不会用部分结果冒充完整报告。

## 使用体验

- 用户只需提供核验问题和想比较的平台；没有指定平台时，立即列出“深知晓、深知晓（深度溯源）、豆包、DeepSeek、通义千问、腾讯元宝”供用户选择，不得漏掉深度溯源，也不要求用户记平台 ID。
- 第一次使用某个平台时打开浏览器，由用户本人完成登录或验证码；会话保存后自动复用。
- 每一步都交付可打开的独立报告：各方答案汇总、各方答案聚合（未核验）、权威核验后的最终答案、各方答案测评报告。
- 每份阶段报告展示后都必须等待用户明确选择继续、修正或结束；最初要求完整执行不能替代逐阶段确认。
- 第二步在概览和逐知识点明细中分别展示直接答案与补充参考；第三步只保留权威最终答案；第四步以一个锁定明细区承接第三步结果并分区测评。
- 第三步只展示最终答案、权威证据和证据边界；第四步只消费已锁定的第三步数据，负责平台裁决、指标与表现对比，不重算第三步。
- 第四步覆盖率只衡量“答没答到”：答到即算覆盖，证据是否充分由证据充分率单独体现。仅当一个知识点在所有平台中都是未回答或证据不足时，才整体移出覆盖率与准确率分母。
- “编造式幻觉”和“证据不足”只作为内部原始裁决状态保留；对用户统一显示为“疑似误导”，移入补充参考风险区并在报告上方突出。每个知识点的长依据默认收起，点击后可看完整原文。
- 来源采集、可信搜索等技术动作失败时原则上停止；若流程仍继续，每一步报告都必须在显著位置提示技术失败及影响，相关主张继续按“疑似误导”呈现。
- 原答案的引用存在性与来源支持性独立保留。`attachedReferenceIndexes` 为当前摘录观察到的引用，`citedReferenceIndexes` 为校验采用的证据；抓取失败或不支持不能写成“平台未附依据”。跨句、跨表格行的引用必须回到对应主张检查，不得借用邻行脚标。

Read the full skill on GitHub

Details

Skill name: fact-check-x-complete
Repository: ASI2030/Fact-Check-X
Path: skills/fact-check-x-complete/SKILL.md
License: Apache-2.0
Repo updated: 19 October 2026
Listed in: karanb192/awesome-claude-skills

FAQ

What does the Fact Check X Complete skill do?

How do I install the Fact Check X Complete skill?

skills CLI (Claude Code, Cursor, Codex and more): npx skills add ASI2030/Fact-Check-X --skill fact-check-x-complete. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Fact Check X Complete popular?

It lives in ASI2030/Fact-Check-X, which has 4 GitHub stars and 0 forks, and is included in 1 curated skill list.

Related skills

Trace My Code

kgohil/trace-my-code

Build and maintain a living trace of a codebase — the domain flow plus the underlying code structure and patterns — as curated Markdown that gives a coding agent the right, always-current context to develop features, follow existing patterns, and refactor safely without re-reading the whole repo. Use this whenever you map or document how a feature/flow works end to end, author or update a DOMAIN/ARCHITECTURE/DATA_FLOW doc or an ADR, record an architecture decision, onboard to or trace an unfamiliar area, link a service's docs to the services it talks to, or refresh docs after code changed — ev

Software DevelopmentIn 1 list

Cra Readiness

Killerz007/cra-readiness-skill

Evidence-backed readiness assessment of a software or hardware product against the EU Cyber Resilience Act, Regulation (EU) 2024/2847, using the pinned Official Journal text, Implementing Regulation (EU) 2025/2392 and the Commission's 2026 guidance. Use when the user mentions the CRA, Cyber Resilience Act, CE marking for software, products with digital elements, Annex I essential requirements, SBOM obligations, vulnerability handling, the 11 September 2026 reporting duty, ENISA single reporting platform, support period, substantial modification, important or critical products, or asks to prepa

Software Development

Anti Gold Plating

TulossSolutions/anti-gold-plating

Prevent gold plating (unapproved extras) through explicit scope, formal validation of improvements and alternatives, and a final compliance summary. Use when creating or modifying a deliverable, code, configuration, or system. A request for analysis or a proposal does not grant authorization for its implementation.

Software Development

Substack API

AnthonyDavidAdams/substack-api-reference

Use Substack's unofficial cookie-authenticated API to read posts, create drafts, publish issues, and manage publications on behalf of a logged-in Substack user. Required when the user asks to post to Substack, schedule a Substack send, mirror content to Substack, or extract data from a Substack account.

Software Development

Ascend Operator Engineering

GTC2080/ascend-operator-skills

Design, implement, review, debug, and tune Huawei Ascend operators with Ascend C or TileLang Ascend. Use for TileLang environment setup, kernel semantics, tiling, UB/DMA/mask/synchronization, ACLNN or AOT integration, precision, MindStudio diagnostics, and trustworthy performance experiments; includes version-scoped lessons from DeepGEMM-Ascend and TileKernels.

Software Development

Skillize Any Lib

pointerliu/skillize-any-libs

Generate SKILL.md for any library by analyzing its source and real-world usage. Supports Rust, Python, JavaScript, Go, C/C++, and more.

Software Development

Browse the most popular Claude skills, all software development skills, or write your own with the skills generator.