What problem does it solve? When a bug appears in the craft-ts demo application, guessing from source code wastes time. This Skill lets you investigate actual runtime evidence: logs captured by the local log server and exposed through the logs MCP bridge, so you can search, filter, and correlate real entries before touching any code. ## Core Features & Use Cases - Log-driven diagnosis: Run logs.stats, then logs.search filtered by level, from, correlationId, clientId, text, or time range to follow one correlated flow across components end to end. - Feature-wide instrumentation: Add a single provideFnWrapper in app.config.ts to log entry, exit, duration, and errors for every craft factory, instead of scattering Console.* calls. - Clean reproductions: Use logs.clear, re-run the exact failing steps, then logs.tail to read the result in emission order. - Troubleshooting empty pipelines: A checklist for when logs.search returns nothing — log server not running, LOG_SERVER_DIR mismatch, raw console.* calls, or buffered entries. - Use Case: A user reports the list page fails after navigation. You run logs.stats, search for error-level entries, take the correlationId of the failure, and retrieve every log of that flow across the route, service, and component hosts. ## Quick Start Use the craft-ts-logs-mcp skill to investigate why the demo application threw an error by searching the captured logs for error-level entries and following their correlationId.
