search-first

Researching existing libraries before coding involves analyzing needs, searching for candidates in parallel, evaluating them against criteria, and deciding to adopt, extend, or build. This structured workflow prevents reinventing the wheel.

The best way to decide whether to adopt, extend, or build a dependency is using a decision matrix that evaluates existing candidates against your feature needs. This matrix guides your engineering workflow by scoring each option.

To avoid reinventing the wheel during new integrations, follow a research-first workflow that locates ready-made solutions and existing patterns before writing any new code. This reduces duplication and accelerates delivery.

You should require a search for existing tools before writing new code during feature discovery, dependency decisions, and new integrations where multiple solutions exist. This prevents wasted engineering effort on duplicated functionality.

A research-first workflow accelerates feature delivery by preventing wasted effort and reducing duplication. By locating ready-made solutions and patterns before coding, teams avoid reinventing the wheel and ship faster.

Evaluating existing libraries involves a structured research workflow of need analysis, parallel search, evaluation against scoring criteria, and a final decision. A researcher-agent applies these criteria to return a recommended action.