turborepo

To configure per-package tasks in a Turborepo monorepo, define the tasks in each individual package.json and register them in the root turbo.json. You can then execute these isolated tasks and leverage parallel execution by running turbo run <task>.

Build caching in multi-package JavaScript repositories works by leveraging Turborepo's dependency graph to avoid rerunning unchanged tasks. This caches task outputs locally and in CI, significantly speeding up workflows by skipping redundant builds across packages.

Yes, you can use per-package turbo.json overrides alongside the root configuration to tailor outputs, environment variables, and caching. This allows framework-specific configurations for tools like Next.js and Vite within your monorepo.

The best way to optimize CI pipelines for affected packages is to use Turborepo's filtering and the --affected flag. This ensures the CI system only builds and tests packages impacted by recent changes rather than running the entire repository task graph.

You need to define task outputs in monorepo package configurations when tasks generate files that downstream packages depend on. Specifying outputs in turbo.json ensures the build caching system correctly stores and restores these generated files for subsequent pipeline steps.

Defining tasks in package.json instead of the repo root enables better parallelization and isolated caching. This per-package approach allows the task orchestrator to map dependencies accurately and skip redundant work across the multi-package repository.