core-sdk

Create and execute Kailash SDK workflows using WorkflowBuilder, nodes, connections, and runtimes.

Updated Apr 2, 2026
One-click install
npx skills add https://github.com/Dchuuuuuu/disease-risk-classifier --skill core-sdk-dchuuuuuu
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: core-sdk
Source: https://github.com/Dchuuuuuu/disease-risk-classifier/tree/main/.claude/skills/01-core-sdk
Command: npx skills add https://github.com/Dchuuuuuu/disease-risk-classifier --skill core-sdk-dchuuuuuu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Building workflow automations with the Kailash SDK requires knowing the correct node patterns, connection syntax, runtime selection, and parameter passing rules, and mistakes like wrong connection arity or using LocalRuntime in Docker cause failures that are hard to diagnose. ## Core Features & Use Cases - Workflow Construction: Build workflows with the string-based WorkflowBuilder API, 4-parameter node configuration, and 4-parameter connection syntax including dot notation for nested data. - Runtime Execution: Choose between LocalRuntime for CLI/scripts and AsyncLocalRuntime for Docker/FastAPI, with async patterns, resource safety, and error handling guidance. - Node Patterns: Use copy-paste templates for PythonCodeNode, AsyncPythonCodeNode, CSV/JSON I/O, HTTP requests, LLM agents, SwitchNode routing, and custom nodes via register_callback. - Use Case: A developer needs to read a CSV, filter rows with Python code, call an external API, and write results to JSON; this Skill provides the exact node, connection, and runtime code to assemble that pipeline correctly. ## Quick Start Ask the AI to create a Kailash workflow that reads a CSV file, processes the data with a PythonCodeNode, and executes it with LocalRuntime.

Frequently Asked Questions about core-sdk

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create a workflow with the Kailash SDK?▼

Create a WorkflowBuilder instance, add nodes with workflow.add_node("NodeName", "id", {config}), connect them with add_connection, then call runtime.execute(workflow.build()). Always call .build() before execution and never call workflow.execute(runtime).

How do I connect nodes in a Kailash workflow?▼

Use the 4-parameter syntax: workflow.add_connection(source_id, source_output, target_id, target_input). For nested fields, use dot notation like "result.metrics.accuracy". The older 3-parameter syntax is deprecated.

Should I use LocalRuntime or AsyncLocalRuntime?▼

Use LocalRuntime for CLI scripts and synchronous contexts. Use AsyncLocalRuntime for Docker and FastAPI deployments, since LocalRuntime in Docker causes hangs. Both return the same (results, run_id) tuple.

How do I pass parameters to Kailash workflow nodes?▼

Use one of three methods: static node configuration in add_node, dynamic workflow connections from other nodes, or runtime parameters passed to execute(). Node-specific runtime parameters are automatically unwrapped and isolated per node.

Does Kailash SDK support cyclic workflows?▼

Cyclic workflows are a planned feature that is not yet implemented in the SDK. Until release, use alternative patterns such as retry logic or manual loops for iterative processing.

Why does my Kailash workflow fail with connection errors?▼

Common causes are wrong port names, mismatched node IDs, or using the deprecated 3-parameter connection syntax. Verify each node's actual input and output ports and match node IDs exactly as declared in add_node.