truthcheck

What This Skill Does

The TruthCheck skill acts as a powerful, autonomous fact-checking assistant within your OpenClaw environment. It leverages the TruthCheck CLI to provide verifiable insights into claims, URLs, and publisher credibility. By performing deep-web searches and optional LLM-based content analysis, the skill evaluates information against a TruthScore scale (0-100), effectively filtering misinformation. It allows users to trace the origins of specific claims, assess the trustworthiness of websites, and maintain an up-to-date repository of publisher reputation data via synchronized databases.

Installation

The recommended installation method for OpenClaw users is via ClawHub, which simplifies dependency management:

npx clawhub install truthcheck

Alternatively, for environments where manual package management is preferred, you can install the underlying Python package directly:

pip install truthcheck

Ensure you have your preferred LLM API keys (e.g., GEMINI_API_KEY) exported in your environment if you intend to use the advanced AI verification features.

Use Cases

1. Misinformation Detection: Quickly verify viral claims or news headlines circulating on social media to determine their objective accuracy.
2. Source Integrity: Investigate whether a specific URL is a reputable news source, a content farm, or a potentially deceptive site.
3. Claim Origin Analysis: When you encounter a specific assertion, use the trace command to see the propagation tree and find the original source of the information.
4. Publisher Vetting: Use the lookup functionality to retrieve historical credibility ratings for publishers before trusting their content.

Example Prompts

1. "OpenClaw, please verify this claim: 'The global average temperature has risen by 1.5 degrees since 1900.' Use the Gemini model for accuracy."
2. "Check if this website is trustworthy and provide a reliability score: https://example-news-site.com."
3. "Can you trace the origin of the rumor regarding the sudden ban of microplastics in industrial manufacturing?"

Tips & Limitations

• Accuracy Enhancement: Always use the --llm flag (e.g., --llm gemini) for complex claims. While the CLI works with default search engines, an LLM provides significantly better contextual analysis.
• Search Performance: Using --search brave often yields superior results compared to the default DuckDuckGo engine for technical or niche topics.
• Time Sensitivity: Verification tasks are intensive; expect a 15-60 second delay for standard checks, and up to 5 minutes for deep traces.
• Environment Security: API keys are never transmitted by the tool; they remain in your local environment variables to authenticate direct calls, ensuring your credentials remain private.
• Sync Regularly: Run truthcheck sync periodically to ensure you are referencing the most current publisher reputation data.