Skip to main content

Nx

Nx Setup

Configure DxCore to parallelize Nx monorepo tasks.

Note

DxCore is not affiliated with, endorsed by, or sponsored by Narwhal Technologies, Inc. (Nrwl). Nx is a trademark of Narwhal Technologies, Inc. See Trademarks & Attribution.

Tip

Nx provides a native task distribution feature — Distribute Task Execution. Check whether it meets your needs before using DxCore’s Nx adapter.

⚠️ Experimental: Nx support is experimental. APIs and behavior may change without notice.

Prerequisites

  • An Nx workspace with project dependencies configured
  • The dxcore binary installed
  • Access to a DxCore coordinator

How It Works

The Nx adapter expects Nx’s task graph JSON piped to stdin. Use nx run-many with --graph=stdout to produce the task graph — this includes per-task metadata like dependency edges and cache configuration. DxCore parses the graph to discover tasks, packages, dependencies, and cacheable flags.

When an agent executes a task, it runs npx nx run <package>:<task> for the assigned work item.

Producing the Task Graph

Pipe the Nx task graph output directly to the dispatcher:

nx run-many --targets=build,test --graph=stdout | dxcore dispatch \

  -c $COORDINATOR_URL -s $SESSION_ID -t $TOKEN -b nx

Warning

Use nx run-many --graph=stdout, not nx graph --file=stdout. The nx graph command only outputs the project dependency graph (no tasks). The nx run-many --graph=stdout command outputs the full task graph including per-task cache booleans that DxCore uses for smart scheduling.

Important

Always pass -b nx to both the dispatcher and agents so the correct adapter is used.

Agent Execution

Each agent must have the repository checked out with dependencies installed. When assigned a task, the agent runs:

npx nx run <package>:<task>

Limitations

Compared to the Turborepo adapter, the Nx adapter has some limitations at plan time:

Feature Turborepo Nx
Task hashes Yes No
Cache status Yes No
Dependencies Yes Yes
Package names Yes Yes

Note

DxCore does not provide its own caching. Nx’s computation cache (local or Nx Cloud) operates independently at execution time — DxCore does not interfere with it and does not have visibility into cache status at plan time.

Next Steps