tlaplus-split-action

To split a TLA+ action, you introduce a new intermediate pc state between the existing action and its successor, updating the original action to transition to this state and creating a new second action guarded on it.

Introducing an intermediate pc state in TLA+ breaks a single action into two sequential transitions, allowing finer-grained control over variable updates and UNCHANGED semantics during model checking.

When splitting a TLA+ action with numbered suffixes, the skill automatically renumbers subsequent actions and updates TypeOk, PCStates, and Next predicates to maintain specification consistency.

When splitting TLA+ actions, the skill preserves or adjusts UNCHANGED semantics by updating the original action to transition to the new intermediate state while maintaining variable consistency across both sequential steps.

Yes, action splitting works with TLA+ specifications using numbered PC states, automatically handling numbered suffixes, renumbering subsequent actions, and providing a ready-to-use ActionX_2 template for the new intermediate action.

The best way to update TypeOk and Next predicates after splitting a TLA+ action is to use automated action splitting, which handles predicate updates alongside pc state insertion and subsequent action renumbering.