search-first

To find existing solutions before coding, use a structured research workflow that performs parallel searches across npm, PyPI, GitHub, and internal references to evaluate viable options. This prevents duplicating effort and rework.

Evaluating dependencies before implementation requires a decision matrix that classifies discovered tools into Adopt, Extend, Build, or Compose categories. This guides your selection based on multi-source survey results from package registries.

The search and decision workflow operates through five steps: NEED ANALYSIS, PARALLEL SEARCH, EVALUATE, DECIDE, and IMPLEMENT. It coordinates a researcher agent to survey multiple sources simultaneously for integration work.

Yes, you can search internal references and GitHub repositories at the same time. The parallel search mechanism surveys npm, PyPI, GitHub, and internal codebases concurrently to identify viable existing solutions.

You should avoid building from scratch when the decision matrix classifies a discovered dependency as Adopt or Extend. This indicates an existing npm, PyPI, or GitHub solution meets your feature requirements without reinventing the wheel.