Personal Research Assistant MCP

import streamlit as st
import json
import pandas as pd
from pathlib import Path
import plotly.express as px

st.title("📊 Metrics Dashboard")

log_dir = "./data/query_logs"

if not Path(log_dir).exists():
    st.warning("No query logs found. Run some searches first!")
else:
    # Load all logs
    logs = []
    for file in Path(log_dir).glob("*.jsonl"):
        with open(file) as f:
            for line in f:
                logs.append(json.loads(line))
    
    if logs:
        df = pd.DataFrame(logs)
        
        # Metrics
        col1, col2, col3 = st.columns(3)
        
        with col1:
            st.metric("Total Queries", len(df))
        
        with col2:
            avg_latency = df['query_time_ms'].mean()
            st.metric("Avg Latency", f"{avg_latency:.1f}ms")
        
        with col3:
            p95_latency = df['query_time_ms'].quantile(0.95)
            st.metric("95th Percentile", f"{p95_latency:.1f}ms")
        
        # Latency histogram
        st.subheader("Query Latency Distribution")
        fig = px.histogram(df, x='query_time_ms', nbins=20, 
                          labels={'query_time_ms': 'Latency (ms)'})
        st.plotly_chart(fig, use_container_width=True)
        
        # Query history table
        st.subheader("Query History")
        display_df = df[['timestamp', 'query', 'query_time_ms', 'results_count', 'top_score']].copy()
        display_df['timestamp'] = pd.to_datetime(display_df['timestamp']).dt.strftime('%Y-%m-%d %H:%M')
        st.dataframe(display_df.sort_values('timestamp', ascending=False), use_container_width=True)
        
    else:
        st.info("No queries logged yet!")