Investigating Code Patterns

This skill provides a structured methodology for navigating, analyzing, and deciphering existing software systems. It is designed for software engineers and technical leads who need to rapidly onboard into new repositories, resolve opaque bugs, or refactor legacy code. By utilizing a rigorous investigative framework—spanning code flow tracing, implementation location, bottleneck identification, and architectural mapping—it transforms ambiguous "how does this work?" questions into evidence-backed technical insights.

• Perform comprehensive code flow tracing to map execution paths from API entry points to data outcomes, ensuring full visibility into decision trees and error handling logic.

• Execute precision-based code location tasks using grep and recursive search patterns to identify main implementation files, supporting modules, and hidden dependencies.

• Diagnose complex performance bottlenecks by applying a structured three-phase approach: locating latency, verifying hypotheses, and implementing evidence-based optimizations for issues like N+1 queries or algorithmic inefficiencies.

• Develop detailed architecture maps that define component boundaries, data flow patterns, and integration points between services or layers.

• Facilitate parallel investigation strategies by coordinating multi-agent workflows to analyze frontend, backend, and integration domains simultaneously.

• Utilize specialized templates to document findings, ensuring that all analysis is anchored to specific file-line references for auditability and team transparency.

• Use this skill to explore code structure, research technology stacks, and answer specific questions about implementation details or architectural rationale.

• This skill is not intended for building new features from scratch; focus on investigation and diagnostic tasks to support later implementation phases.

• Start by reading project documentation, including product requirements, feature specs, system designs, and API contracts (e.g., swagger/openapi/yaml) before diving into deep code analysis.

• Always prioritize citing specific file paths and line numbers as evidence to avoid speculative conclusions.

• Employ direct tools like grep, read, and glob for pattern discovery, and use specialized agent orchestration when tackling multi-service or full-stack performance investigations.