Dagpipe

DagPipe is a reliability layer for AI workflows, focused on making LLM (Large Language Model) pipelines robust, crash-proof, and cost-efficient for developers, non-coders, and agent/app users. It prevents cascade failures in multi-node workflows by validating and checkpointing each node's state, ensuring uninterrupted, safe recovery from any node failure. With features like live model routing, semantic output assertions, and support for provider-agnostic model functions, DagPipe empowers the building, orchestration, and safe deployment of AI-driven pipelines both as a Python package and through MCP (Model Context Protocol) integration.