Optimizely DXP MCP Server

# 📊 Log Analysis & Download Capabilities ## What We've Built ### 1. **Log Download from Azure Storage** (`download_logs`) Downloads Application Insights logs from Azure Storage containers. #### Log Types Available: - **Application Logs** - Application-specific logging - **Web Logs** - IIS/web server logs - **CloudFlare Logs** - CDN logs (if configured) - **All** - Downloads all available log types #### Time-Based Filtering: ```bash # Simple day-based filtering claude "download logs from last 7 days" # daysBack: 7 claude "download logs from last 24 hours" # daysBack: 1 # Date range filtering claude "download logs from August 20 to 25" # startDate/endDate # Specific date filtering claude "download logs from 2025/08/24" # dateFilter ``` #### Timezone Support (NEW): ```bash # Automatic timezone detection claude "download logs from August 27" # Automatically converts your local August 27 to UTC # Explicit timezone specification claude "download logs from August 27 timezone EST" claude "download logs from August 27 timezone PST" claude "download logs from August 27 timezone UTC" # Timezone offset claude "download logs from August 27 timezone -5" # EST claude "download logs from August 27 timezone +2" # CEST ``` **Important**: Azure logs are stored in UTC, but the system now automatically: - Detects your local timezone - Converts your requested dates to UTC - Expands date ranges to cover the full day in your timezone - Shows timezone adjustment in descriptions #### Features: - **90 days of logs available** (Azure Storage retention) - **Automatic container detection** (insights-logs-* pattern) - **Size preview** before downloading - **Progress tracking** during download - **Configurable download paths** per project - **Stale log detection** (NEW) - Alerts if Production logs are >30 days old - **Timezone-aware filtering** (NEW) - Automatically handles UTC conversion ### 2. **Intelligent Log Analysis** (`analyze_logs`) AI-powered log analysis that downloads logs temporarily, analyzes them, then cleans up. #### Pattern Detection: - **Errors**: ERROR, FATAL, Exception, Failed, crashes - **Warnings**: WARN, deprecated, timeout - **Security**: unauthorized, forbidden, authentication failed - **Performance**: slow, timeout, latency issues - **Database**: SQL errors, deadlocks, connection issues - **HTTP Errors**: 404, 500, 503, etc. #### Analysis Output: ```bash claude "analyze logs from production" # Returns: # 📊 Log Analysis Report # - Total Lines: 45,231 # - Errors: 142 (0.31%) # - Warnings: 523 # - Unique Errors: 27 # # 🔍 Pattern Detection # - HTTP Errors: 89 occurrences # - Database issues: 12 occurrences # - Security events: 3 occurrences # # 🎯 Recommendations # - High error rate detected # - Review authentication failures # - Database connection pooling issues ``` #### Natural Language Queries: ```bash # Find specific issues claude "find errors in production logs from last hour" claude "show performance issues in staging logs" claude "detect security events in last 24 hours" # Get statistics claude "show log statistics for production" claude "analyze error trends over last week" ``` ## How It Works ### Log Download Process: 1. **Connects to Azure Storage** using project credentials 2. **Lists available containers** (insights-logs-applicationlogs, etc.) 3. **Filters by date** if specified 4. **Shows preview** with file count and total size 5. **Downloads to configured path** (default: ./logs) 6. **Tracks progress** with real-time updates ### Log Analysis Process: 1. **Downloads logs to temp directory** 2. **Parses and analyzes** log content 3. **Detects patterns** using regex matching 4. **Generates insights** based on findings 5. **Creates recommendations** 6. **Cleans up temp files** automatically ## Configuration ### Download Paths: ```bash # Set project-specific log download path claude "set log download path to ./logs/project-name" # Environment variables (in .mcp.json) "PROJECT_NAME": "logPath=/path/to/logs" ``` ### Default Behaviors: - **Environment**: Production (unless specified) - **Time Range**: Last 24 hours (daysBack: 1) - **Log Type**: Application logs - **Download Path**: ./logs or project-specific path ## Examples ### Basic Log Download: ```bash # Download today's application logs from production claude "download logs" # Download last week's logs from staging claude "download logs from staging for last 7 days" # Download specific date range claude "download logs from August 20 to August 25" ``` ### Log Analysis: ```bash # Quick error check claude "analyze logs for errors" # Performance analysis claude "find performance issues in logs from last hour" # Security audit claude "check for security events in production logs" ``` ### Advanced Filtering: ```bash # Specific log type and timeframe claude "download web logs from last 3 days" # Multiple filters claude "analyze application logs from staging environment for last 48 hours" ``` ## Limitations 1. **90-day retention** - Azure Storage keeps logs for 90 days 2. **Large file handling** - Very large log files may take time to download/analyze 3. **Temp storage** - Analysis requires temporary local storage 4. **Pattern matching** - Regex-based, may not catch all variations ## Best Practices 1. **Preview First**: Use preview mode to check size before downloading 2. **Time Filtering**: Always specify timeframe to avoid large downloads 3. **Regular Analysis**: Run periodic analyses to catch issues early 4. **Clean Up**: Downloaded logs accumulate - clean old files regularly ## Future Enhancements (Not Yet Implemented) - Real-time log streaming - Custom pattern definitions - Log aggregation across environments - Automated alerting on patterns - Historical trend analysis - Export to monitoring platforms