CrewAI Setup: Scientific Review Board

Initialize a CrewAI project. Define at least three agents with distinct roles for a 'Scientific Review Board': a 'Factual Verifier', a 'Consistency Checker', and an 'AI Slop Detector'. Assign each agent a goal and a backstory. Configure them to use Claude Opus 4.1 as their primary LLM. Provide the Python code for agent setup.

```python
from crewai import Agent, Task, Crew, Process
from langchain_anthropic import ChatAnthropic # Use specific Langchain integration for Claude

# Initialize Claude Opus 4.1 LLM
claude_opus_llm = ChatAnthropic(model="claude-3-opus-20240229", temperature=0.2, anthropic_api_key="YOUR_ANTHROPIC_API_KEY")

# Factual Verifier Agent
factual_verifier = Agent(
    role='Factual Verifier',
    goal='Verify the factual accuracy of claims in scientific texts against established knowledge.',
    backstory='An meticulous researcher with access to vast scientific databases, ensuring every statement is evidence-backed.',
    verbose=True,
    allow_delegation=False,
    llm=claude_opus_llm
)

# Consistency Checker Agent
consistency_checker = Agent(
    role='Consistency Checker',
    goal='Identify logical inconsistencies, contradictions, and methodological flaws within a scientific document.',
    backstory='A sharp-eyed critic who ensures coherence and methodological rigor in scientific discourse.',
    verbose=True,
    allow_delegation=True,
    llm=claude_opus_llm
)

# AI Slop Detector Agent
ai_slop_detector = Agent(
    role='AI Slop Detector',
    goal='Detect generic phrasing, vague claims, repetitive patterns, and other hallmarks of low-quality, potentially AI-generated text.',
    backstory='A seasoned editor with an uncanny ability to spot superficial writing and lack of original thought, especially from generative models.',
    verbose=True,
    allow_delegation=False,
    llm=claude_opus_llm
)

# (Tasks and Crew definition will follow)
```