LogStitch

Product Introduction

1. Definition: LogStitch is a native macOS desktop application designed as a specialized AWS Lambda log analysis and debugging tool. It operates as a local-first observability platform that parses, stitches, and analyzes raw CloudWatch log streams.
2. Core Value Proposition: LogStitch exists to solve the fundamental problem of scattered and interleaved AWS Lambda logs by automatically reconstructing coherent, request-based invocations. Its primary value is providing fast root cause analysis for serverless failures, eliminating manual correlation across CloudWatch streams, and offering integrated performance analytics without requiring external dashboards or subscriptions.

Main Features

1. Log Stitching & Invocation Grouping: LogStitch reads the unique request ID stamped on every CloudWatch log line and automatically groups all related lines from across multiple concurrent executions, functions, accounts, and regions into a single, legible invocation card. It parses structured JSON, flags cold starts, and calculates duration and memory usage for each request, transforming a chaotic timestamp-sorted log river into discrete, actionable stories. This is achieved through local processing in milliseconds.
2. Cross-Function Correlation & Swim-Lane Timeline: The software provides a correlation view that traces a single user action across multiple Lambda function hops. By searching a request ID or correlation header, it generates a swim-lane timeline showing propagation latency between services, the origin of an error, and the downstream blast radius. It can infer correlations using temporal proximity when explicit IDs are missing and highlight broken chains.
3. Integrated Performance Analytics & Detection: LogStitch includes a built-in analytics dashboard that calculates p99/p95 latency trends, cold-start distributions, memory utilization, and a monthly cost projection directly from locally stored data. Its detection engine automatically clusters recurring errors into patterns with lifecycle states (e.g., worsening, improving) and uses Z-score statistical analysis to surface anomalies in duration, error rate, cold-start frequency, and cost, alerting users to deviations like duration regressions or error spikes.
4. Local MCP Server for AI Integration: The application ships a local Model Context Protocol (MCP) server bound to 127.0.0.1. This allows AI assistants like Claude Code to securely query the local LogStitch SQLite database via functions like search_logs, get_cost_projection, and list_patterns. This integration enables AI-powered debugging without exposing AWS credentials or log data to external networks, keeping all interactions machine-local.
5. Local-First Architecture & Privacy: All data is stored in a local SQLite database on the user's Mac. AWS credentials and OAuth tokens (for Jira, GitHub) are stored securely in the macOS Keychain. The application makes direct API calls to AWS from the user's machine, operates fully offline after initial sync, and transmits no log data or telemetry to any external LogStitch backend, ensuring strict data privacy and control.

Problems Solved

1. Pain Point: The primary problem is debugging failures in AWS Lambda using CloudWatch Logs, where log lines from multiple concurrent executions are interleaved, making it extremely time-consuming to reconstruct the sequence of events for a single failed request. This leads to extended Mean Time to Resolution (MTTR) for production issues.
2. Target Audience: The core users are AWS Backend Developers, DevOps/SRE Engineers, Serverless Architects, and Technical Leads managing microservices or event-driven architectures built on AWS Lambda. It is essential for teams practicing on-call rotations and individuals responsible for performance optimization and cost control of serverless applications.
3. Use Cases: LogStitch is critical for post-mortem analysis after a production outage, debugging intermittent cold-start-related failures, identifying latency bottlenecks in multi-service invocations, validating memory right-sizing for cost savings, and proactively monitoring error pattern trends before they cause major incidents. It is also used for auditing and optimizing AWS Lambda costs and correlating issues with Jira or GitHub tickets.

Unique Advantages

1. Differentiation vs. CloudWatch Logs Insights: While CloudWatch Logs Insights is a powerful query tool over raw log streams within the AWS console, LogStitch offers a fundamentally different paradigm. It provides a persistent, local, invocation-centric view rather than ephemeral queries. It automatically correlates logs across disparate AWS accounts and regions and delivers integrated analytics and AI querying via a local MCP server without ever requiring the log data to leave the user's machine.
2. Key Innovation: The defining innovation is the local-first, invocation-stitching engine combined with a local MCP server. This unique architecture creates a secure, private observability layer where the intelligence is on the edge (the user's Mac). The MCP server acts as a secure bridge, enabling advanced AI tools to leverage structured log analysis capabilities without compromising security, a feature absent in traditional log management SaaS or console tools.

Frequently Asked Questions (FAQ)

1. Is LogStitch a secure way to let AI analyze my AWS logs? Yes. LogStitch's local MCP server runs exclusively on your machine (127.0.0.1) and queries its own local SQLite database. Your AWS credentials never leave the macOS Keychain, and log data is never sent to an external LogStitch server. This provides a secure, private method for AI-assisted debugging.
2. How is this different from using CloudWatch Logs Insights or Metric Filters? LogStitch complements those tools by providing a persistent, visual, and correlated context that Insights lacks. It automatically stitches logs by request ID across accounts/regions and correlates invocations into a timeline, whereas Insights requires writing specific query language (Analytics) for each ad-hoc search. LogStitch also includes built-in statistical anomaly detection and cost analysis not native to the console.
3. Does LogStitch work with all AWS Lambda runtimes and regions? Yes. LogStitch works with any AWS Lambda runtime (Node.js, Python, Java, Go, etc.) as it parses the standard CloudWatch log format that includes the request ID. It supports all AWS regions and is designed to correlate invocations across multiple functions, accounts, and regions in a single view.
4. What happens to my data if I stop using LogStitch? All your data is stored in a local SQLite file on your Mac. You retain full ownership and can back it up, move it, or delete it at any time. The application also supports export to JSON or CSV. There is no cloud dependency for your historical log data.
5. Is the one-time purchase price for all future versions? The pricing is a one-time purchase for a perpetual license to the version you buy. It includes all future updates within that major version series (e.g., v1.x). Major version upgrades may be sold separately, following standard macOS App Store practices.