# dify-for-dsl: A Community Collection of Dify Workflow DSL Scripts

> dify-for-dsl is an unlicensed community collection of 50+ Dify workflow DSL files in YAML format, covering AI image generation, document processing, video creation, text-to-speech, data queries, and more. The README is in Chinese and requires Dify 0.8.0 or later for import.

**wwwzhouhui/dify-for-dsl** — 本项目是基于dify开源项目实现的dsl工作流脚本合集

- Repository: https://github.com/wwwzhouhui/dify-for-dsl
- Stars: 3,875 · Forks: 742
- Language: Python
- License: not declared
- Published: 2026-09-23 · Updated: 2026-09-23 · Language: en
- Canonical page: https://hysenlabs.com/projects/wwwzhouhui-dify-for-dsl

## What dify-for-dsl Is and Who It Covers

Dify is an open-source LLM application platform that lets users build AI workflows through a visual interface. Those workflows can be exported and imported as DSL files, which are YAML-formatted descriptions of the workflow nodes, connections, and configurations. dify-for-dsl is a community collection of those DSL files, maintained primarily by the author wwwzhouhui and a small number of contributors.

The README, written in Chinese, describes the project as: "a collection of DSL workflow scripts implemented based on the Dify open-source project, sharing some useful Dify workflows, good for personal use and learning." The instruction is to use Dify 0.8.0 or later for importing the files.

The target users are Dify practitioners who want to learn from working workflow examples, adapt existing patterns for their own use cases, or bypass the time investment of building common workflows from scratch. The collection skews toward Chinese-language use cases (Chinese medicine, Chinese exam grading, Chinese invoice processing) but includes universal workflows for document translation, image generation, and data queries.

## Repository Organization: How to Navigate the Collection

The repository is organized into five functional directories:

- `dsl/` holds the .yml workflow files, which are the primary content of the repository. Each file is a complete importable Dify workflow.
- `py/` contains Python helper scripts used alongside some workflows.
- `mcp/` holds MCP (Model Context Protocol) server configurations for workflows that connect to external tools via MCP.
- `plugin/` contains plugin files, including several custom plugins developed by the author.
- `assets/` holds screenshots and image previews of the workflow visual displays.

The README's DSL table is the main navigation tool. Each row lists the workflow file name, a screenshot of the workflow canvas, the technologies used, the last update date, the author, and the minimum compatible Dify version. The table spans Dify versions from 0.0.1 in the earliest entry (周易大师.yml, a divination chatflow from 2025-03-26) to 1.9.2 in the most recent entries.

Since the README is in Chinese, non-Chinese readers navigating the collection should focus on the workflow file names, which contain enough English keywords (the model names, API services, and Dify component names) to identify the workflow's purpose.

## Importing a DSL Workflow into Dify

The README describes a four-step import process:

1. Open Dify.
2. Go to the Create Application area and select Import DSL.
3. Select the .yml file from the dsl/ directory.
4. Confirm creation.

This is the only technical prerequisite documented in the README. There is no build step, no dependency installation, and no Python environment needed for the import itself. The workflows run inside the Dify instance once imported.

Some workflows depend on specific plugins or external APIs that must be configured separately within Dify before the workflow can run. For example, workflows using the Nano-Banana image generation plugin require that plugin to be installed and configured in the Dify plugin marketplace. Workflows that call external APIs via HTTP request nodes require valid API credentials. The README table's "technologies used" column identifies these dependencies for each workflow.

For Dify version compatibility, the table specifies the minimum version for each workflow. Newer workflows in the collection target Dify 1.6.0 through 1.9.2, while older entries support versions as low as 0.0.1. Running a 1.9.2-targeting workflow on a Dify 0.8.0 installation may fail because newer node types or features are not available in the older version.

## Domain Coverage Across the Workflow Collection

The collection spans a wide range of AI application domains. Image generation workflows integrate Nano Banana2 for image creation with negative prompts and style selection, Qwen-Image for free text-to-image and image-to-image conversion, and Gemini 2.0 Flash for image generation via condition branching.

Video generation workflows target Sora2 (via a custom plugin developed by the author), the Jiemeng (即梦) platform via HTTP request nodes and reverse-engineered API calls, and Doubao video generators. Text-to-speech workflows include a free EdgeTTS plugin integration.

Document processing is a strong category: PDF invoice batch recognition to Excel output using multi-modal LLM and iteration nodes; contract review for software development contracts with WeChat Work notification; multi-language PDF translation preserving the original format using Agent and MCP-Server; math exercise generation and error tracking for students using SQL Execute and ECharts chart generation.

Data and query workflows include text-to-SQL generation with knowledge base retrieval, Chinese company profile visualization with 16 chart types using MCP chart servers, student grade lookup with ECharts, and train ticket queries via the 12306 MCP integration. RSS news aggregation pulls from eight platforms and sends email summaries via SMTP. The collection also includes a Mermaid-based architecture diagram generator, a prompt generator chatflow, and a mind map creation workflow.

## Custom Plugins Authored by the Repository Maintainer

Several workflows in this collection depend on custom Dify plugins developed by the repository's primary author, wwwzhouhui. These are documented in the workflow table's technology column and include:

- Nano-Banana plugin: used in image generation workflows that target the Nano Banana2AI service
- qwen-image plugin: used in free Qwen-Image text-to-image and image-to-image workflows
- free_edgetts plugin: used in text-to-speech workflows for free TTS via Edge TTS
- Sora2 plugin: used in workflows that generate video through the Sora2 service

These plugins are referenced as being in the plugin/ directory of the repository. For any workflow that lists one of these as a dependency, the plugin must be installed in the Dify instance first. Since these are not official Dify marketplace plugins, their installation and maintenance fall outside the standard Dify update cycle.

Some workflows also use third-party tools authored by others, such as the rookie_rss tool for RSS aggregation, a PDF-to-PNG converter, and community MCP servers. The README credits these dependencies in the workflow table but does not provide installation steps.

## Limitations: No License, Reverse-Engineered APIs, and Version Fragmentation

The repository declares no license. GitHub identifies the license as unknown. This means any legal use beyond personal, private experimentation is unclear. Teams that want to use or adapt these workflows in commercial products should seek clarification from the author or treat the code as all-rights-reserved until a license is added.

Several workflows use reverse-engineered API endpoints for services that do not offer official APIs, such as the Jiemeng video generation platform. The README explicitly labels these as reverse-engineered (逆向接口, 逆向API). Reverse-engineered endpoints break when the target service changes its interface, and there is no guarantee of continued support. Workflows built on these endpoints may stop working without notice.

The Dify version matrix creates a fragmentation problem. With workflows spanning from Dify 0.0.1 to 1.9.2, a user on any given Dify version will find some workflows incompatible. The table lists the minimum version but not the maximum; it is possible that workflows targeting old Dify node structures may not function correctly on very recent Dify releases. Checking compatibility before import is necessary.

The documentation is entirely in Chinese. File names, the README, and workflow descriptions are written for a Chinese-speaking audience.

## Dify's Official Template Library as an Alternative

Dify maintains its own template library and application marketplace as part of the platform. These are official, curated workflows that are tested against current Dify releases and supported by the Dify team. The official templates represent a narrower but more stable selection compared to this community collection.

The difference in approach is curation versus variety. Official Dify templates go through a review process and are intended to demonstrate the platform's canonical patterns. dify-for-dsl is a personal collection that prioritizes breadth and experimentation, including workflows that use unpublished plugins, reverse-engineered services, and cutting-edge model integrations that may not be stable.

For teams new to Dify who want to understand how workflows are structured without risk of hitting unstable dependencies, the official templates are the safer starting point. For practitioners who already know Dify and want a catalogue of community-tested patterns covering niche use cases like Chinese invoice OCR, 12306 train ticket queries, or Sora2 video generation, this collection provides workflows that the official library does not cover.

## Conclusion

Dify users who want a library of ready-made workflows to import and adapt for image generation, document OCR, video creation, SQL queries, TTS, and news aggregation should explore dify-for-dsl. Two things to verify before adopting any specific workflow: the Dify version requirement listed in the README's table (workflows range from version 0.0.1 to 1.9.2, so older installs may not support newer entries), and whether the APIs and plugins each workflow depends on are still available (some use reverse-engineered endpoints that may change without notice). The repository has no declared license; review the terms before any commercial or redistributed use. The last push to this repository was on 2026-09-27.

## FAQ

### What Dify version is required to use these workflows?

The README states that Dify 0.8.0 or later is required for importing workflows. Each workflow in the DSL table also lists its own minimum compatible Dify version, which ranges from 0.0.1 for the earliest entries to 1.9.2 for the most recent. Check the version column in the README table before importing a specific workflow.

### How do I import a DSL workflow from this repository into Dify?

Open Dify, go to the Create Application area, select Import DSL, choose the .yml file from the dsl/ directory, and confirm creation. Some workflows require specific plugins or API credentials to be configured in your Dify instance before the workflow can run; the technologies column in the README table lists each workflow's dependencies.

### Do the workflows in dify-for-dsl require specific plugins or third-party services?

Many workflows depend on plugins or external APIs. Some plugins (Nano-Banana, qwen-image, free_edgetts, Sora2) were developed by the repository author and are in the plugin/ directory. Others use third-party tools like rookie_rss and community MCP servers. Some workflows use reverse-engineered API endpoints that are not officially supported by the target services.

## Sources

- [Issues](https://github.com/wwwzhouhui/dify-for-dsl/issues)
- [README](https://github.com/wwwzhouhui/dify-for-dsl/blob/main/README.md)
- [wwwzhouhui/dify-for-dsl on GitHub](https://github.com/wwwzhouhui/dify-for-dsl)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/wwwzhouhui-dify-for-dsl