← 返回 Skills 市场
1001
总下载
0
收藏
6
当前安装
1
版本数
在 OpenClaw 中安装
/install fact-check
功能描述
Verify claims, statements, and information against reliable sources. Use when user asks to fact-check, verify, confirm, validate, or check the accuracy of an...
使用说明 (SKILL.md)
Fact Check
Verify information through systematic source verification.
Workflow
- Identify the claim - Extract the specific statement to verify
- Search sources - Query reliable sources (news outlets, academic papers, official databases)
- Cross-reference - Check multiple independent sources
- Evaluate credibility - Assess source reliability and potential biases
- Deliver verdict - Provide clear conclusion with evidence
Claim Types
| Type | Approach | Trusted Sources |
|---|---|---|
| Statistics | Find original study/data | Government databases, peer-reviewed journals |
| Quotes | Locate primary source | Official transcripts, news archives |
| News events | Verify via multiple outlets | Major news organizations |
| Historical facts | Consult academic sources | Encyclopedias, scholarly publications |
| Product/company info | Check official sources | Company websites, SEC filings |
Source Hierarchy
Tier 1 (Highest):
- Peer-reviewed journals
- Government databases/official statistics
- Court records, legal documents
Tier 2:
- Major news outlets (Reuters, AP, BBC, etc.)
- Established academic institutions
- Official company/organization statements
Tier 3:
- Social media (verify via other sources)
- Blogs, personal websites
- Unverified reports
Output Format
Claim: [exact statement]
Verdict: [TRUE / MOSTLY TRUE / UNVERIFIABLE / MOSTLY FALSE / FALSE]
Evidence:
- Source 1: [details]
- Source 2: [details]
Notes: [context, caveats, or additional context]
Constraints
- Never state certainty beyond what evidence supports
- Distinguish between "unverified" and "false"
- Provide sources, don't just claim verification
- Flag opinions vs facts clearly
- Update verdict if new evidence emerges
安全使用建议
This skill appears coherent and low-risk, but consider these practical points before enabling: (1) it assumes the agent can access external sources — ensure you trust the agent's browsing/integration tools and their destination domains, (2) avoid sending sensitive or confidential statements for fact-checking because the skill may query third-party services, (3) if you need strict provenance, require the skill to cite primary sources with URLs and dates, and (4) if you want to limit source bias, configure an allowlist/denylist of domains or require specific databases (e.g., government or academic repositories). If any of those controls matter to you, request the skill author add explicit source/tool constraints or require specific credentials rather than leaving source selection open-ended.
功能分析
Type: OpenClaw Skill
Name: fact-check
Version: 1.0.0
The skill bundle contains only metadata and documentation (SKILL.md) outlining a standard workflow for fact-checking information. There is no executable code, no network activity, and no evidence of malicious instructions or prompt injection attempts.
能力评估
Purpose & Capability
Name, description, and SKILL.md align: instructions describe extracting claims, searching reliable sources, cross-referencing, and producing a verdict. The skill does not request unrelated credentials, binaries, or config paths.
Instruction Scope
Instructions are appropriately scoped to fact‑checking (identify claim, search sources, evaluate credibility, produce verdict). They are somewhat high‑level and assume the agent can query external sources (web, news, journals) but do not instruct the agent to read local files or access unrelated system state. Because source selection is left open, the agent's actual behavior will depend on the platform's browsing/API tools.
Install Mechanism
No install spec and no code files (instruction-only), so nothing is written to disk or downloaded during install — lowest install risk.
Credentials
No environment variables, credentials, or config paths are required. The declared needs are proportionate to the stated purpose. If the agent uses network tools or third‑party APIs to implement searches, those tools (and any credentials they need) are not specified by this skill.
Persistence & Privilege
always:false and user-invocable:true (defaults). The skill does not request persistent/system-wide privileges or configuration changes. Autonomous invocation is permitted by default but is not combined with other risky privileges.
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 fact-check - 安装完成后，直接呼叫该 Skill 的名称或使用
/fact-check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
v1.0.0
- Initial release of the fact-check skill.
- Provides systematic workflow for verifying statements using reliable sources.
- Supports fact-checking of statistics, quotes, news events, historical facts, and company info.
- Assigns verdicts (TRUE, MOSTLY TRUE, UNVERIFIABLE, MOSTLY FALSE, FALSE) with cited evidence.
- Distinguishes between facts, unverified claims, and opinions with clear output format.
元数据
常见问题
Fact Check 是什么？
Verify claims, statements, and information against reliable sources. Use when user asks to fact-check, verify, confirm, validate, or check the accuracy of an... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 1001 次。
如何安装 Fact Check？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install fact-check」即可一键安装，无需额外配置。
Fact Check 是免费的吗？
是的，Fact Check 完全免费（开源免费），可自由下载、安装和使用。
Fact Check 支持哪些平台？
Fact Check 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Fact Check？
由 古 晖（@webguhui）开发并维护，当前版本 v1.0.0。
推荐 Skills