dify-dsl-generator_skill

Overview

This skill is a professional Dify workflow DSL/YML generator that converts plain business requirements into complete, import-ready Dify configuration files. It supports all major node types, automatic edge wiring, plugin dependency detection, and produces YAML that follows Dify 0.3.0 DSL rules. Use it to speed up workflow authoring, reduce manual errors, and enforce consistent structure and IDs.

How this skill works

Provide a short specification of the desired workflow (purpose, inputs, processing steps, outputs, optional plugins) and the skill generates a full app/dependencies/workflow YAML file. It designs start nodes, LLM prompts, code/http/tool/condition nodes, variable mappings, edges, positions, and recommended model parameters. The generator validates IDs, variable reference formats, and typical YAML constraints so output is import-ready for Dify.

When to use it

• You need a ready-to-import Dify DSL for a new chatflow, workflow, or agent.
• You want to convert a business process description into an executable workflow quickly.
• You need consistent, validated YAML that follows Dify 0.3.0 conventions.
• You require multi-node flows with LLM, code, HTTP, tool, and conditional logic.
• You want automatic plugin/dependency configuration for external integrations.

Best practices

• Describe inputs, processing steps, and expected outputs clearly and concisely.
• Specify required plugins or external APIs to ensure dependencies are included.
• Provide example inputs or schemas for data-driven nodes (SQL, JSON, files).
• Review suggested prompt templates and adjust system/user prompts for domain needs.
• Use unique, timestamp-based node IDs and keep node variable names semantic.

Example use cases

• Image OCR: file upload start -> vision-enabled LLM -> answer node returning recognized text.
• HTML generator: text input -> LLM to produce HTML -> parameter-extractor -> md_exporter tool -> answer with file URL.
• Data query: user question -> LLM generates SQL -> HTTP request to DB API -> code node formats results -> LLM builds chart HTML -> answer displays chart.
• AI agent: task input -> LLM planner -> parallel tool calls -> variable aggregator -> LLM summarizer -> final answer.