How to Search npm Packages by Keyword

npm's website is excellent for interactive discovery, but repeatable ecosystem research needs structured inputs, preserved metrics, and an exportable result. The npm Package Scraper searches the public registry and returns one normalized row for each unique package.

Write a focused registry query

Begin with the capability a developer would search for and keep the initial result count small.

{
  "queries": ["web scraping"],
  "packageNames": [],
  "maxResultsPerQuery": 25
}

Search rows include the package name, current version, description, keywords, publisher, maintainers, weekly and monthly download estimates, dependent count, npm score components, and project links. The query is retained so results from several searches can be traced back to their discovery context.

Review beyond the ranking

Registry rank is a useful discovery signal, not a dependency recommendation. Inspect the repository, release history, license, documentation, security advisories, and compatibility with your runtime before adoption. Download counts can be influenced by automated builds and transitive dependencies.

Deduplication uses the package name, so one package matching several queries appears once in a run. If query-level overlap itself matters, run searches separately or store the raw query membership before merging.

For a recurring landscape report, save the input as an Apify Task and retain timestamped datasets. Compare versions and metrics by package name. A missing row may mean it fell below your result cap rather than that it was unpublished. Keep the query, cap, and collection time with every comparison, and treat a failed run as a monitoring gap instead of an empty ecosystem.