Getting Started

Getting Started

Installation

pip install rvrb-verify

Provider setup

rvrb-verify requires an LLM provider. Set one of:

Provider	Environment Variable	Get it at
Qwen (default)	DASHSCOPE_API_KEY	Alibaba Cloud
OpenAI	OPENAI_API_KEY	platform.openai.com
Local	None	Ollama or vLLM running locally

First verification

CLI

# Basic verification
export DASHSCOPE_API_KEY="sk-..."
rvrb-verify "The sky is blue"

# With specific strategy
rvrb-verify "Water boils at 100°C" --strategy fact-check

# JSON output
rvrb-verify "Python was created in 1991" --json

# Research strategy
rvrb-verify "Quantum computing breaks Moore's law" --strategy research

# Save to file
rvrb-verify "The Earth is flat" --output verdict.txt

Python

from rvrb_verify import verify

# Basic usage
verdict = verify("The sky is green")

print(f"Verdict: {verdict.verdict.value}")  # "false"
print(f"Confidence: {verdict.confidence:.1%}")  # 0.95
print(f"Summary: {verdict.summary}")

# With strategy
verdict = verify("Water boils at 100°C", strategy="fact-check")

# With model override
verdict = verify("Claim text", model="qwen3-coder-plus")

Strategies

Strategy	When to use	Example claims
fact-check	General factual claims	"The sky is blue", "Water boils at 100°C"
legal	Legal/statutory claims	"GDPR requires consent for cookies"
research	Academic/scientific claims	"Quantum computing breaks Moore's law"

Output formats

Format	CLI flag	Description
Plain text	(default)	Verdict, confidence, summary
JSON	--json	Full verdict with evidence and sources

What's next

• CLI Reference — all flags and options
• Python API — programmatic usage
• Strategies — deep dive into each strategy
• MCP Server — integrate with AI agents