Pals Not Working at Base: Fix the Actual Layer

1. Remove the known version problem

The current v1.0.3 patch specifically addresses deployed Pals stopping combat and failing to assist with base work. Update the host and every client before rebuilding the base around behavior that may already be fixed.

On a dedicated server, verify the server binary was updated too. A current client does not repair an older host process.

2. Prove the job is valid

1. 01

   Check the station's required work suitability and the Pal's current suitability.

2. 02

   Confirm the station has input materials, power where required and output storage space.

3. 03

   Assign the Pal directly instead of relying on automatic priority.

4. 04

   Watch for a task message; if it immediately changes jobs, remove competing stations for the test.

3. Test access, not appearance

Palworld 1.0 can begin some jobs even when the worker cannot physically stand where expected, so visual distance alone is no longer a complete pathing test. Palbox crowding, narrow doors, low ceilings, stacked stations and large bodies can still interrupt real uptime.

Move one station onto flat open ground near the Palbox and assign one medium or small Pal. If that works, the worker is healthy and the production layout—not the recipe—is the problem.

4. Restore the worker loop

1. 01

   Fill the Feed Box and make sure food is reachable.

2. 02

   Check SAN, injuries and status effects; replace the Pal temporarily while it recovers.

3. 03

   Provide beds and temperature-appropriate facilities inside the base boundary.

4. 04

   Put the Pal in the Palbox and redeploy it after the station is cleared.

5. Isolate a server-only problem

If a clean one-Pal test works in single player but not on the server, check server logs, memory pressure, worker-count changes and mods. Reproduce the issue without mods before changing dozens of base objects.

Record whether the Pal is idle, switching tasks, starving, stuck at the Palbox or apparently working without output. Those are different failures and should not share one generic fix.