core-sdk

Build workflows using WorkflowBuilder to add nodes, define connections, and pass parameters. Create a builder instance, add nodes with configuration, call .build(), then execute via LocalRuntime or AsyncLocalRuntime to run your automation.

LocalRuntime executes workflows synchronously, blocking until completion. AsyncLocalRuntime runs workflows asynchronously, allowing non-blocking execution. Choose based on whether your automation needs immediate results or can operate in the background.

Parameters flow through node connections using 4-parameter connection syntax. Define outputs from one node, connect them to another node's inputs, and access results via the runtime's returned results dictionary keyed by node name.

Yes, PythonCodeNode lets you embed custom Python code directly in workflows. Pass code as a configuration parameter, and the node executes it during workflow runtime, returning results to downstream nodes.

Kailash supports error handling through node-level validation and cyclic patterns for retry logic. Design workflows to capture exceptions, route failures to handler nodes, and use async patterns for fault tolerance across distributed execution contexts.

Yes, Kailash includes MCP integration for protocol-based communication and enterprise-ready monitoring, validation, and debugging features. Auto runtime-detection adjusts execution context, supporting 110+ nodes and 29 configuration parameters for scalable automation.