Build Policy Interpretation with Claude Opus 4.1

Using the Claude Agents SDK, build the 'Compliance Policy Agent'. This agent should receive transcribed content (from the 'Content Ingestor/Analyzer Agent') and a list of policy rules. The agent must use Claude Opus 4.1's advanced reasoning to determine if the content violates any rules, identify the specific violation type, and provide a clear explanation. Configure the agent to be verbose, demonstrating its thought process in making a decision. Show how you define the agent and its interaction loop.

```python
from anthropic import Anthropic
from claude_agents.api import Agent, AgentBuilder

# Initialize Anthropic client
client = Anthropic(api_key=os.environ.get("ANTHROPIC_API_KEY"))

def check_compliance(content: str, policy_rules: list[str]) -> dict:
    # Agent logic will go here. Example prompt for Claude.
    prompt = f"Given the content: '{content}' and these policy rules: {policy_rules}. " \
             "Determine if any rule is violated. If so, identify the violation type, explain why, and suggest an action." 
    
    response = client.messages.create(
        model="claude-3-opus-20240229", # Use Claude Opus 4.1
        max_tokens=1000,
        messages=[
            {"role": "user", "content": prompt}
        ]
    )
    # Parse response to structured dict
    return {"decision": "violating", "violation_type": "...", "explanation": "...", "recommended_action": "..."}

# Example of how you'd define this as an agent skill
# compliance_agent_skill = AgentBuilder(client=client).tool(
#     name="check_content_compliance",
#     description="Checks content against a set of policy rules."
# )(check_compliance)
```