VeritySwarm — AI Truth Verification Swarm

The strongest truth-verification engine on the internet.

verityswarm.com · AI fact-checking · Multi-agent verification · Confidence scores · Source citations

---

What is VeritySwarm?

VeritySwarm deploys a coordinated swarm of 6 specialized AI agents to verify any claim — gathering evidence, stress-testing it adversarially, detecting bias, auditing sources, and placing it in historical context before a final synthesis delivers a definitive verdict.

No spin. No hedging. Just transparent, structured truth.

---

How It Works

Claim Input
    │
    ▼
┌─────────────────────────────── Round 1: Independent Analysis ──────────────────────────────┐
│  🔍 Researcher      Evidence mining, fact gathering, source assembly                        │
│  ⚔️  Skeptic         Adversarial challenge, counter-evidence, gap analysis                  │
│  ⚖️  Bias Detector   Political/commercial/ideological framing analysis                      │
│  🔗 Source Cross-Checker   Credibility audit, inter-source consistency check               │
│  📜 Context Historian      Historical precedent, claim origin, pattern matching             │
└────────────────────────────────────────────────────────────────────────────────────────────┘
    │
    ▼ (each agent reads all peers' Round 1 output)
    │
┌─────────────────────────────── Round 2: Cross-Pollinating Refinement ──────────────────────┐
│  All 5 agents refine their analysis in light of peer findings                              │
│  Confidence scores shift. New red flags emerge. Weak arguments are challenged.             │
└────────────────────────────────────────────────────────────────────────────────────────────┘
    │
    ▼
┌─────────────────────────────── Final Synthesis (Grok → Gemini → Claude) ───────────────────┐
│  🏛️  Synthesizer    Weighs all evidence across both rounds                                  │
│                    Issues definitive verdict with full reasoning chain                     │
└────────────────────────────────────────────────────────────────────────────────────────────┘
    │
    ▼
Verity Report: Verdict · Confidence Score · Sources · Red Flags · Key Insights · PDF

---

Verity Report Output

Every report includes:

Field	Description
Verdict	TRUE FALSE PARTIALLY TRUE MISLEADING UNVERIFIED
Confidence Score	0–100 integer, weighted across all agents and rounds
Summary	2–3 sentences, plain language, no hedging
Synthesis	Multi-paragraph integrated reasoning
Sources	Consolidated citations from all agents
Red Flags	SNAKE_CASE tags: SINGLE_SOURCE, RECURRING_MYTH, POLITICAL_SPIN...
Key Insights	Synthesized takeaways any sharp reader needs to know
Reasoning Chain	Full per-agent analysis (Round 1 and Round 2)
PDF Export	Dark-themed professional PDF via reportlab

---

Quick Start

# Install
pip install anthropic openai google-generativeai reportlab

# Set API keys in .env
ANTHROPIC_API_KEY=your_key
GEMINI_API_KEY=your_key
XAI_API_KEY=your_key  # optional (Grok synthesis)

# Run
python agents/swarm.py "Humans only use 10 percent of their brain"

# Options
python agents/swarm.py --rounds 1 --no-pdf "The moon landing was faked"
python agents/swarm.py --quiet "Wind turbines cause cancer" > report.txt

---

Architecture

verityswarm/
├── agents/
│   ├── swarm.py          # Core swarm engine (v2)
│   └── __init__.py
├── legal/
│   ├── TERMS_OF_SERVICE.md
│   ├── PRIVACY_POLICY.md
│   ├── TRADE_SECRETS.md
│   └── NDA_TEMPLATE.md
├── cloudflare/
│   ├── setup_tunnel.bat  # One-click tunnel setup
│   └── config_template.yml
├── reports/              # Timestamped JSON + PDF history
└── index.html            # Landing page (Tailwind, dark mode)

---

Agent Configuration

Each agent outputs strict JSON:

{
  "analysis":     "detailed prose analysis",
  "confidence":   0-100,
  "sources":      ["source 1", "source 2"],
  "red_flags":    ["SNAKE_CASE_FLAGS"],
  "key_insights": ["concise insight 1", "concise insight 2"]
}

The Synthesizer aggregates all agents across both rounds and outputs:

{
  "verdict":      "FALSE",
  "confidence":   94,
  "summary":      "Plain-language 2-3 sentence verdict.",
  "synthesis":    "Full integrated reasoning...",
  "sources":      ["..."],
  "red_flags":    ["PERSISTENT_MYTH", "DEBUNKED_PREVIOUSLY"],
  "key_insights": ["..."]
}

---

Pricing

Plan	Price	Limit
Free	$0	3 verifications/day
Pro	$9/mo	Unlimited + API + PDF
Enterprise	Custom	Bulk + white-label

Get Pro at verityswarm.com

---

Keywords

AI fact checking · automated fact verification · truth verification AI · claim verification tool · misinformation detection · AI skeptic · bias detection tool · source credibility checker · multi-agent AI · swarm intelligence fact check · LLM fact checker · Grok verification · Claude fact check · AI truth engine

---

Legal

Proprietary. All rights reserved. © 2026 AcKccreate. See TRADE_SECRETS.md for IP protection details. Contact: hello@verityswarm.com