CodeVetter · Repo Unpack · Feature catalog
nrwl/nx
Nx exposes its primary capabilities through a centrally registered CLI command surface. Task execution supports individual projects, multiple projects, affected selection, custom commands, caching, and several output modes. Workspace maintenance, generation, cloud operations, and agent integration are additional command families. The graph browser and documentation provide separate interactive and reference surfaces; this catalog groups the inspected features without asserting exhaustive coverage of every plugin.
- Users can run project:target:configuration or infix target/project syntax; both handlers call runOne, and --skip-nx-cache is documented on the run command.
- Affected execution can select tasks through computeAffectedTasks when enabled and targets are supplied, or select affected projects through filterAffected; graph requests retain a separate visualization branch.
- Generation resolves plugin schemas and implementations, combines options with workspace defaults, supports help and dry-run, and delegates non-Nx generators through the Angular compatibility adapter.
- The command table registers workspace setup and maintenance families including add, init, import, list, migrate, release, repair, report, reset, format, show, sync, watch, and run-many.
- Cloud and agent command families include login/logout, record, start-ci-run, start-nx-agents, start-agent, stop-all-agents, fix-ci, apply-locally, download-cloud-client, configure-ai-agents, MCP, and shell completion.
- Caching supports local artifacts, Nx Cloud, optional S3/shared-filesystem/GCS/Azure packages, and a self-hosted HTTP provider; HTTP remote caching is explicitly unavailable under WASM.
- The graph browser exposes projects, tasks, tasks/all, and project-details routes; task loading accepts target/project selections, returns an empty graph for insufficient selection, and caches combinations by workspace.
- Core executors provide noop, run-commands, and run-script; run-commands supports parallel execution and readiness markers, while task output can use summary, dynamic/static renderers, or the native TUI.