Verify Claims

Verify factual claims in content using deep research methodology. Can be run standalone or after NCI analysis.

Usage

/decipon:verify <URL or text content>
/decipon:verify --claims "claim1" "claim2" "claim3"

Arguments

• $ARGUMENTS: Content to analyze OR specific claims to verify
  • Content: URL, file path, or text
  • --claims: Verify specific claims directly

Purpose

Fact-check key claims using deep research:

• Extract factual assertions from content
• Research each claim using multiple sources
• Score source reliability (1-100 confidence)
• Track contradictions and agreements
• Report verification status with evidence

Workflow

1. Extract Claims (if content provided)

   • Identify factual assertions
   • Prioritize by impact (Critical/High/Medium/Low)
   • Focus on verifiable claims (not opinions)

2. Research Each Claim

   • Apply deep research methodology
   • Search for supporting/contradicting evidence
   • Evaluate source quality
   • Track agreements and contradictions

3. Evaluate Sources

   • Score confidence (1-100)
   • Note source type and potential bias
   • Check for corroboration

4. Generate Report

   • Verification status per claim
   • Source citations with confidence
   • Overall credibility assessment

Verification Statuses

Status	Meaning
VERIFIED	2+ high-confidence sources support claim
PARTIALLY VERIFIED	Core claim true but with caveats
UNVERIFIED	Cannot find supporting evidence
CONTRADICTED	High-confidence sources refute claim

Output Format

═══════════════════════════════════════════════════
  CLAIM VERIFICATION REPORT
═══════════════════════════════════════════════════

SUMMARY
───────
Claims Analyzed: [N]
✓ Verified:      [N]
~ Partial:       [N]
? Unverified:    [N]
✗ Contradicted:  [N]

DETAILED FINDINGS
─────────────────

CLAIM 1: "[Statement]"
Status: [STATUS]
Confidence: [%]

Sources:
  [1] example.com (Confidence: 85) - Supports
  [2] reuters.com (Confidence: 90) - Supports

Finding: [What research revealed]

───────────────────────────────────────────────────

CLAIM 2: "[Statement]"
[Continue for each claim...]

═══════════════════════════════════════════════════

METHODOLOGY
Applied deep research verification protocol
Consulted [N] sources total
Source confidence scoring: 1-100 scale

RECOMMENDATIONS
[Any suggested follow-up verification]

═══════════════════════════════════════════════════

Integration with NCI Analysis

After /decipon:analyze

If NCI analysis shows elevated manipulation risk (score > 40), verification is recommended:

/decipon:analyze https://example.com/article
# Returns NCI score of 55 with key claims identified

/decipon:verify https://example.com/article
# Verifies claims and may adjust manipulation assessment

Automatic Triggers

Verification is automatically suggested when:

• NCI Score > 40
• Authority Issues category > 3
• Suspicious Timing category > 3
• Cherry-Picking category > 3

Score Adjustment

After verification, NCI scores may be adjusted:

• Verified claims: May reduce manipulation scores
• Contradicted claims: Increase manipulation scores
• Report shows original vs adjusted scores

Example Invocations

Verify content from URL:

/decipon:verify https://example.com/news-article

Verify specific claims:

/decipon:verify --claims "The study showed 90% improvement" "Experts unanimously agree"

After NCI analysis:

/decipon:analyze <content>
# If score elevated:
/decipon:verify <content>

Deep Research Integration

Uses the deep research skill for:

• Source evaluation and confidence scoring
• Contradiction handling protocol
• Search patterns for different source types
• Quality iteration until confidence threshold met

See skills/deep-research/SKILL.md for full methodology.

Source Confidence Guidelines

Source Type	Confidence Range
Peer-reviewed research	85-100
Official documentation	85-95
Government/institutional	75-90
Major news (Reuters, AP)	70-85
Established newspapers	60-80
Industry publications	50-75
Blogs/forums	20-50
Anonymous/undated	5-20

Comparison with Other Commands

Command	Focus	Depth
/decipon:score	Quick manipulation score	Light
/decipon:analyze	Full pattern analysis	Thorough
/decipon:verify	Claim fact-checking	Deep research
/decipon:report	Formal documentation	Complete

Best Practice Workflow

1. Quick triage: /decipon:score - Get initial manipulation score
2. Full analysis: /decipon:analyze - If score > 20
3. Fact-check: /decipon:verify - If score > 40 or claims seem dubious
4. Document: /decipon:report - For formal records