Agent Fact Check Verify

This skill addresses the difficulty of quickly verifying the authenticity of information. In today's landscape of social media and information overload, users often encounter claims that need fact-checking, but manually searching multiple sources is time-consuming and prone to missing key information. This skill integrates multiple channels—including government sources, official media, fact-checking websites, X (Twitter), and Reddit—to automatically perform rigorous cross-verification. It employs an internal 100-point scoring system (not publicly disclosed) and ultimately provides a clear judgment: correct, false, partially correct, or insufficient evidence.

Usage is simple: just make a verification request in natural language, such as "Check if this news is true" or "Verify this message." The skill automatically breaks down the core claim, executes searches according to a preset query budget allocation (Tavily 50%, Twitter 40%, Reddit 10%), and enforces Tavily priority. If API limits are encountered, it automatically downgrades to backup searches. It then calculates a credibility score through internal rules and finally outputs a four-part conclusion: a brief answer, the actual situation, the conclusion, and relevant links (up to 5). The entire process requires no user intervention, delivering an integrated and reliable answer directly.

It is well-suited for teams or individuals who need to quickly verify information authenticity. This is especially useful for content creators (to avoid spreading rumors), academic researchers (to ensure reliable citations), social media managers (to monitor the truthfulness of public sentiment), journalists (for fact-checking), and general users (to discern online information). For users who have already configured a Tavily API key and require high-frequency verification, this skill can significantly improve efficiency and reduce the risk of being misled.

It is recommended to use this tool uniformly when verifying high-risk information such as public safety, healthcare, and financial scams. Note that it primarily relies on publicly available information and cannot cover paywalled or non-public content. For predictive information (e.g., future events), it does not make true/false judgments but only compiles existing verifiable situations. Additionally, the skill includes a "misjudgment reduction insurance" mechanism to avoid overly strict false judgments due to breaking news tone or non-critical details, though core factual errors will still be directly judged as false.