research-first

Root-cause tracing requires verifying issues by searching current documentation and tracing problems end-to-end through actual source code to reach ≥95% confidence before proposing changes. This approach prevents fixing symptoms instead of underlying faults.

Researching state-of-the-art approaches involves deploying parallel sub-agents to gather diverse evidence across multiple online sources simultaneously. This synthesizes current official documentation and best practices to inform technical decisions and architecture.

Plan research before coding fixes by systematically searching official documentation and literature first, then tracing the reported issue through the codebase. This ensures fixes are grounded in current guidance rather than assumptions.

Yes, parallel sub-agents can be deployed to explore multiple sources simultaneously during issue diagnosis. They gather diverse evidence and synthesize findings to validate root-cause tracing before you implement any code changes.

You need to search official documentation whenever investigating technical debt, architecting solutions, or diagnosing unknown issues. Validating findings against current guidance before tracing code paths prevents implementing outdated or incorrect fixes.

Code fixes fail when they target symptoms instead of the root cause. Achieving ≥95% confidence through end-to-end source code tracing and validating findings against current documentation ensures the actual fault is addressed.