import os
from mcp_agents.claude_client import get_anthropic_client
class Summarizer:
"""Agent responsible for text summarization with multiple detail levels using Claude"""
def __init__(self):
self.client = get_anthropic_client()
self.model = os.getenv('ANTHROPIC_MODEL', 'claude-3-5-sonnet-20241022')
# Define summary configurations for different detail levels
self.summary_configs = {
'quick': {
'max_tokens': 300,
'system_prompt': """You are an expert summarizer. Create a quick, concise summary with 2-3 key bullet points.
Focus only on the most essential information. Keep it brief and actionable.""",
'user_prompt': "Provide a quick summary with 2-3 key bullet points of the most important information:",
'description': "Quick overview with 2-3 key points"
},
'standard': {
'max_tokens': 600,
'system_prompt': """You are an expert summarizer. Create a balanced summary with 4-5 main topics.
Include key details and context while maintaining readability. This is the default level.""",
'user_prompt': "Provide a standard summary covering the main topics and key details:",
'description': "Balanced summary with main topics and details"
},
'detailed': {
'max_tokens': 1200,
'system_prompt': """You are an expert summarizer. Create a comprehensive analysis with detailed insights.
Organize information into 6+ sections with context, implications, and deeper analysis.""",
'user_prompt': "Provide a detailed, comprehensive summary with analysis, context, and implications:",
'description': "Comprehensive analysis with context and insights"
}
}
def summarize(self, text: str, detail_level: str = 'standard') -> str:
"""
Summarize text with specified detail level using Claude
Args:
text (str): Text to summarize
detail_level (str): Summary level - 'quick', 'standard', or 'detailed'
Returns:
str: Generated summary
"""
# Validate and set default detail level
if detail_level not in self.summary_configs:
detail_level = 'standard'
config = self.summary_configs[detail_level]
# Handle empty or very short text
if not text or len(text.strip()) < 50:
return "Text is too short to summarize meaningfully."
# Truncate very long text to avoid token limits
max_input_chars = 50000 # Claude can handle more than GPT
if len(text) > max_input_chars:
text = text[:max_input_chars] + "..."
try:
response = self.client.messages.create(
model=self.model,
max_tokens=config['max_tokens'],
system=config['system_prompt'],
messages=[
{
"role": "user",
"content": f"{config['user_prompt']}\n\n{text}"
}
]
)
summary = response.content[0].text.strip()
# Add summary level indicator
level_indicator = f"[{detail_level.upper()} SUMMARY]\n\n"
return level_indicator + summary
except Exception as e:
error_msg = f"Error generating {detail_level} summary: {str(e)}"
print(f"❌ Summarizer error: {error_msg}")
return error_msg
def get_available_levels(self) -> dict:
"""
Get available summary levels and their descriptions
Returns:
dict: Dictionary of level names and descriptions
"""
return {level: config['description'] for level, config in self.summary_configs.items()}
def get_level_info(self, level: str) -> dict:
"""
Get information about a specific summary level
Args:
level (str): Summary level name
Returns:
dict: Level configuration info
"""
if level in self.summary_configs:
config = self.summary_configs[level]
return {
'level': level,
'description': config['description'],
'max_tokens': config['max_tokens']
}
return None
