Usage #
Use this prompt to verify claims, statements, or assertions from articles, presentations, or any public content.
Prompt #
Verify the following claim using open-source intelligence:
**Claim:** [CLAIM]
**Source:** [WHERE THE CLAIM APPEARED]
**Context:** [BRIEF CONTEXT]
## Your task:
### 1. Decompose the claim
Break it into individual verifiable assertions.
### 2. Verify each assertion
For each sub-claim:
- Search for primary sources (academic papers, official statistics, original research)
- Search for corroborating secondary sources
- Note any contradicting evidence
- Rate confidence: high / medium / low
### 3. Evaluate the source
- Author credentials and affiliations
- Publication venue reputation
- Methodology used (if applicable)
- Potential conflicts of interest
- Date of publication
### 4. Check for bias
- Framing analysis (what's emphasized vs. omitted)
- Language analysis (objective vs. persuasive)
- Funding or commercial interests
- Selection bias in data or examples
### 5. Provide context
- Historical background relevant to the claim
- Expert consensus on the topic
- Related research or events
- Cultural or regional factors
### 6. Verdict
Rate the overall claim:
- TRUE: Fully supported by evidence
- MOSTLY TRUE: Core claim accurate, minor inaccuracies
- MIXED: Some parts true, some false or misleading
- MOSTLY FALSE: Core claim inaccurate despite some true elements
- FALSE: Contradicted by evidence
- UNVERIFIABLE: Insufficient evidence to determine
### 7. Confidence level
- High / Medium / Low with justification
## Output format:
Structured markdown with numbered sources. Include a "Limitations" section.
Maintain neutral, analytical tone throughout.Variables #
[CLAIM]— the exact claim to verify[WHERE THE CLAIM APPEARED]— article URL, presentation, etc.[BRIEF CONTEXT]— why this matters, what prompted the check
Notes #
- Based on methodology:
methodology/fact-verification.md - For multi-agent use, run 4 instances with different focus areas (see methodology)
- Always include limitations section