DagPipe

Crash-proof LLM pipeline orchestration with DAG-based execution, checkpoint recovery, and cognitive model routing.

Provides a reliability layer for LLM workflows with crash recovery, schema validation, and cost-aware model routing. Pipelines are defined as directed acyclic graphs where each node's output is independently validated and checkpointed. Failed runs resume from the last successful node without losing progress. Includes a smart model router that selects models based on task complexity, a live model registry, and semantic output assertions.