investigate

To investigate an error traceback without modifying source code, use this skill to spawn parallel subagents that examine primary source files, dependencies, tests, and architecture. It identifies the root cause and data flow, then writes a structured markdown report to temp/investigate/ with recommendations.

Analyzing a complex module and its architectural data flow is best handled by deploying parallel subagents to explore the codebase from multiple angles. This approach identifies patterns, maps data flow, and documents findings in a structured markdown report for future reference.

Parallel code analysis works by guiding multiple subagents simultaneously to investigate errors, modules, tests, and architecture. This multi-angle exploration safely examines the codebase without making changes, resulting in a comprehensive markdown report detailing root causes and prevention patterns.

Yes, you can use this codebase investigation approach for understanding dependencies and tests. The parallel subagents explicitly examine these dimensions alongside primary source files and architecture to provide a complete picture of the issue.

The primary limitation of this read-only investigation is that it strictly enforces no source code modifications. It only diagnoses issues and writes a markdown report to temp/investigate/ detailing root causes and recommendations, requiring separate actions to actually implement fixes.

To start a deep code investigation, provide a detailed request describing the bug, error, or architectural question. The system will automatically spawn parallel subagents to explore the codebase and generate a comprehensive structured markdown report.