search-first

To research existing libraries before coding, automate a pre-coding research sprint to scan npm, PyPI, MCP servers, and GitHub, evaluating candidate patterns to ensure you reuse ready-made solutions instead of reinventing the wheel.

A decision matrix for choosing software libraries evaluates candidates with a consistent scoring framework, guiding your team to clearly adopt an existing tool as-is, extend or wrap it, or build a custom solution.

To compare cross-language tools for a dependency upgrade, run parallel research across multiple package registries and GitHub to surface candidates, then score them using a consistent evaluation matrix.

The best way to organize research artifacts for toolchain decisions is to systematically capture evaluation scores and notes during your research sprint, aligning the outputs with planner agents to drive implementation plans.

You should build custom code instead of adopting an existing library when your evaluation matrix indicates that available candidates fail to meet your project's scoring criteria for feature fit and integration requirements.