Discover. Extend. Build

Open‑Box Deals Aggregator is a Dify tool powered by the TinyFish Web Agent. It searches multiple major retailers for open‑box, refurbished, and clearance deals in real time, then returns **clean, structured results** (e.g., product name, price, condition, and URL) that you can use inside your Dify workflows. Use it to: - Compare deals across retailers quickly - Build price‑tracking or deal‑alert workflows - Power shopping, sourcing, or research assistants with agent‑ready outputs

This template continuously monitors specific folders in Google Drive and automatically indexes new or modified files into a Qdrant vector database. When the Google Drive Change Trigger reports a change, the workflow downloads the file, runs OCR for images or extracts text from documents, splits the text into manageable chunks and generates embeddings. Those embeddings are upserted into a Qdrant collection for later retrieval. This allows you to build and refresh a knowledge base from documents stored in Google Drive without manual intervention.

This workflow implements an agentic retrieval‑augmented‑generation (RAG) pipeline for legal research. The Agent node inspects the user’s question and selects the most appropriate action: search pre‑built Qdrant collections (e.g. CUAD mini, Iowa housing law, Wyoming housing law) or perform a Google web search. Search results are summarised and cited, and the agent returns a natural‑language answer while following strict instructions (e.g. no legal advice). The pipeline is designed for questions about contract clauses and housing regulations and demonstrates a multi‑tool, context‑aware RAG agent

Researches a market or industry in depth using Tavily web search. You provide an industry, optional focus angle, and time horizon. The workflow automatically identifies 4–6 emerging trends, searches for real-world evidence on each in parallel, scores every trend on momentum, market size, feasibility, and threat resistance (each 1–5), then synthesizes a ranked opportunity report with a comparative scoring matrix and strategic recommendations.

Give it an industry, an optional focus angle, and a time horizon — it automatically identifies 4–6 specific emerging trends, searches for real-world evidence on each in parallel using Tavily, and scores every trend on four dimensions (momentum, market size, feasibility, and disruption threat). A code node computes an opportunity score and ranks the trends, then a final LLM synthesizes a polished briefing with an executive summary, a scored opportunity matrix table, and strategic recommendations on where to act.

Describe your company, what you offer, and your target industry — the workflow finds potential clients for you. It generates 5–8 targeted discovery searches, runs them in parallel via Tavily to surface real candidate companies, deduplicates the results, then qualifies each company with two searches (company profile + live buying signals like funding, hiring, and growth news). A scoring model rates each lead on fit and intent, routes them into Tier A (hot), Tier B (warm), or Tier C (nurture), and drafts a tailored outreach message for every lead — personalized emails for Tier A, shorter notes f

Scan AI Agent Skills and MCP projects before installation or deployment. This template lets users provide one scan source, such as an uploaded .zip package, a Git repository URL, or a remote MCP URL. The SkillWard plugin sends the source to the hosted SkillWard Sentinel backend for static analysis, LLM-based safety evaluation, and optional runtime sandbox verification, then returns a concise security summary and structured scan report.

Feed in a rendered video and get back a soundtrack generated from the cut itself — video-to-music, not a text prompt. Licensed, safe for commercial use (terms apply). Requires a Sonilo API key (sonilo.com).

Smart chatbot with built-in memory for long conversations.

Convert business cards into structured digital data with a basic workflow.

Digital receptionist that sorts and routes users inquiries to specialized AI agents based on their intent.

Find out if your statement is a fact or a myth.

Build a smart chatbot that answers questions based on your own knowledge base.

Stay on top of any topic with real-time news summaries.

Paste a company website URL and this workflow will generate a basic structured market research report.

Upload a file and get a clean and organized summary.

Automatic UTM link generation with Code node.

Get a clean, AI-powered summary of the latest technology stories from The New York Times.

Send your creative brief, get a high-quality draft, and review before publishing.

24/7 digital security for your website.

Open‑Box Deals Aggregator is a Dify tool powered by the TinyFish Web Agent. It searches multiple major retailers for open‑box, refurbished, and clearance deals in real time, then returns **clean, structured results** (e.g., product name, price, condition, and URL) that you can use inside your Dify workflows. Use it to: - Compare deals across retailers quickly - Build price‑tracking or deal‑alert workflows - Power shopping, sourcing, or research assistants with agent‑ready outputs

Automatically generate production-ready Dify workflow YAML from plain text requirements. Skip the tedious drag-and-drop design — just describe your agent, and this tool outputs a fully functional, importable workflow. Supports all core nodes including LLM, Code, If-Else, Iteration, and more. The generated workflow can be further edited in Dify for iteration and optimization.

The workflow accepts a user-uploaded document and a preferred target language, then uses GPT-4o to translate the full document content into that language. It includes a file size guard that routes invalid uploads to an error output before any translation attempt is made. The flow is: Start → IF/ELSE (size check) → Doc Extractor → LLM Translation → Output. If the file fails the size check, it short-circuits to a separate error output.

A semantic search pipeline that combines MongoDB Atlas Vector Search with VoyageAI embeddings and reranking for high-precision RAG retrieval. User input is embedded using voyage-3.5, searched against a MongoDB Atlas vector index, and reranked with rerank-2.5 to surface the most relevant documents. Results are formatted via a template node for downstream LLM use. Suitable for knowledge bases, document Q&A, and AI agent memory retrieval.

This workflow is designed to automate and standardize first-pass email triage for support teams as ticket volume grows. Instead of relying on manual judgment to decide who should handle each incoming case, it creates a structured routing process that is fast, consistent, and auditable.

Generate data-driven SEO content briefs by scraping Google SERP with Bright Data, analyzing top-ranking pages with OpenAI GPT-4o, and producing editor-ready briefs in Markdown.

Input 6 Amazon product URLs (yours + 5 competitors), output a Markdown competitive analysis report. Fetches metadata + Amazon's AI review summary (customer_says) via Bright Data, then runs GPT-4o-mini in 5 parallel evaluations: praise, complaints, positioning, improvements, differentiation. Why customer_says-first: Amazon's AI distills thousands of reviews into ~250 chars per product, native-language. Higher signal-to-noise than raw reviews. Pay-per-run, open YAML, amazon.co.jp validated. v0.2: customer_says-based. v0.3 plans full Reviews. Requires Bright Data + OpenAI keys. MIT.

DeepResearch uses loop variables and agent nodes to iteratively explore the web, identify knowledge gaps, and synthesize findings into a structured report. The workflow leverages tools like Tavily Search to collect relevant web results and Deepseek Reasoner to analyze and summarize them, delivering comprehensive answers to complex questions.

Start by entering a question, search using Tavily Search, crawl web content with Jina Reader, and generate summaries to get a structured results table.

Deep Research workflow built with iteration, reasons, and sources until the answer is complete.

Researches a market or industry in depth using Tavily web search. You provide an industry, optional focus angle, and time horizon. The workflow automatically identifies 4–6 emerging trends, searches for real-world evidence on each in parallel, scores every trend on momentum, market size, feasibility, and threat resistance (each 1–5), then synthesizes a ranked opportunity report with a comparative scoring matrix and strategic recommendations.

Give it an industry, an optional focus angle, and a time horizon — it automatically identifies 4–6 specific emerging trends, searches for real-world evidence on each in parallel using Tavily, and scores every trend on four dimensions (momentum, market size, feasibility, and disruption threat). A code node computes an opportunity score and ranks the trends, then a final LLM synthesizes a polished briefing with an executive summary, a scored opportunity matrix table, and strategic recommendations on where to act.

This workflow automatically turns long-form text into a visually consistent presentation deck. It first extracts and structures the input content into slide-level outlines, then generates a styled presentation cover and slide images in a unified visual theme.Each slide is rendered individually using an image-capable LLM, ensuring consistent branding, typography, and layout across the entire deck. The result is a ready-to-use presentation generated directly from raw text, without manual slide design.

Paste a company website URL and this workflow will generate a basic structured market research report.

Automatic UTM link generation with Code node.

Generate data-driven SEO content briefs by scraping Google SERP with Bright Data, analyzing top-ranking pages with OpenAI GPT-4o, and producing editor-ready briefs in Markdown.

This template turns a draft, reference files, and a target platform list into tailored social posts for each platform. It helps marketers, creators, and content teams generate consistent platform-specific content with the right tone and language. The workflow also extracts information from uploaded documents and images, then formats the final output for easy review.

The provided images illustrate a sophisticated workflow architecture. This structure utilizes a modular design to manage complex logic, breaking down intricate processes into independent, collaborative nodes that optimize information flow. The system serves as a logical topology that emphasizes the necessity of hierarchical processing and dynamic scheduling when handling complex tasks. As highlighted in image_caae84.png, the requirement to draft a brief description for a template represents the abstraction of these complex technical systems into clear, actionable value propositions.

An enterprise-grade workflow designed to automatically transform raw input (drafts, text, images, or documents) into tailored, publication-ready social media posts for multiple platforms (e.g., Red/Xiaohongshu, WeChat Moments, Twitter, LinkedIn) simultaneously. Key Highlights: - Intelligent Input Cleaning: Built-in Parameter Extractor standardizes platform inputs. - Robust Exception Handling: Smart IF/ELSE routing prevents invalid token consumption. - Multimodal Processing: Automatically separates and parses images and text documents.

An AI-powered SEO workflow that transforms keywords, competitor insights, and market research into strategic content opportunities and optimized article outlines.

Author：Zhiqiang Yang A scheduled weather notification workflow that sends a daily weather forecast email at a fixed time. It retrieves weather data for a specified city using the OpenWeather plugin, formats the forecast into a clear, human-readable email with an LLM, and automatically delivers it via Gmail. Suitable for daily planning, personal alerts, or location-based notification use cases.

DeepResearch uses loop variables and agent nodes to iteratively explore the web, identify knowledge gaps, and synthesize findings into a structured report. The workflow leverages tools like Tavily Search to collect relevant web results and Deepseek Reasoner to analyze and summarize them, delivering comprehensive answers to complex questions.

Write SQL from natural language by pasting in your schema with the request. Please describe your query requirements in natural language and select the target database type.

I am a YouTube Channel Data Analysis Copilot, I am here to provide expert data analysis tailored to your needs.

Deep Research workflow built with iteration, reasons, and sources until the answer is complete.

Upload PDFs or documents and get high-quality translations in seconds.

An AI that actually gets to know you. This template uses memory logic to extract and store facts, preferences, and details about you from your conversation in real-time.

This assistant uses multi-turn logic to collect patient's information like age, symptoms, and gender, while identifies missing information and asks follow-up questions.

Transform messy meeting transcripts into structured notes instantly.

Perform batch sentiment analysis on text and return JSON output including sentiment classification and score.

Welcome to your personalized Investment Analysis Copilot service, where we delve into the depths of stock analysis to provide you with comprehensive insights.

I can answer your questions related to strategic marketing.

Code Interpreter helps developers to understand code and discover errors. It clarifies the syntax and semantics of the code step by step.

Start by entering a question, search using Tavily Search, crawl web content with Jina Reader, and generate summaries to get a structured results table.

Basic Workflow Template, a chatbot capable of identifying intents alongside with a knowledge base.

Basic Workflow Template. A chatbot with a knowledge base.

Providing translation capabilities in multiple programming languages, translating the user's input code into the programming language they need.

Author: Suliu Qin An automated workflow that scours the web for the latest AI news and product updates based on your keywords. It then intelligently summarizes the findings and generates a beautifully formatted, ready-to-share HTML daily digest. Stay ahead of the curve effortlessly.

This instructional workflow demonstrates how to build a multi‑step pipeline in Dify. It accepts a body of text, extracts all standalone questions using a parameter‑extraction node, loops through each question, retrieves relevant knowledge from a dataset, and uses an agent to draft an answer with citations. Finally, the answers are formatted into a structured reply. The template is intended for educational purposes to help new users understand loops, iterations, knowledge retrieval and agent nodes.

This workflow automatically turns long-form text into a visually consistent presentation deck. It first extracts and structures the input content into slide-level outlines, then generates a styled presentation cover and slide images in a unified visual theme.Each slide is rendered individually using an image-capable LLM, ensuring consistent branding, typography, and layout across the entire deck. The result is a ready-to-use presentation generated directly from raw text, without manual slide design.

This workflow acts as an intelligent ticket‑handling system. It receives ticket details via a webhook, extracts key information, classifies the ticket type (e.g. return, account issue, product question), retrieves relevant articles from a knowledge base and generates a response. It can also send notifications to team channels (e.g., Slack or Discord) and keep track of ticket status. The template is suitable for helpdesk automation and customer service triage.

A daily scheduled workflow that checks whether required Docker images are missing from a target registry and triggers synchronization only when needed. On a Daily Trigger, it refreshes metadata, fetches the latest version, and validates whether any image variants are missing. If missing variants are found, an Agent parses the JSON, extracts namespace, name, and version, and calls the “Run Github Workflow” tool to trigger a GitHub Actions sync job. Finally, it updates the cron endpoint with the current version. Must be used with the companion “Github Action” tool template.

This workflow automatically turns long-form text into a visually consistent presentation deck. It first extracts and structures the input content into slide-level outlines, then generates a styled presentation cover and slide images in a unified visual theme.Each slide is rendered individually using an image-capable LLM, ensuring consistent branding, typography, and layout across the entire deck. The result is a ready-to-use presentation generated directly from raw text, without manual slide design.

This workflow automates the handling of inbound customer emails. Given a customer’s name, email subject and body, it classifies the query into predefined categories (e.g. product inquiry, refund request), retrieves relevant knowledge from your dataset, and uses an agent to compose a personalised reply that addresses the customer’s concerns. It streamlines customer support and ensures consistent responses.

Author：Tianwei Wu A daily digest of the top-xranked Hugging Face research papers, written for non-researchers. Each paper is explained in plain English, highlighting its core idea, its real contribution, and who it is actually useful for. The goal is to help you quickly decide which of today’s most important papers are worth reading, without jargon, hype, or unnecessary detail.

This workflow is a daily briefing generator. A schedule trigger fires at a preset time each day, triggering a search for trending AI/technology news via the Tavily Search plugin. The results are summarised by an LLM into a concise brief and delivered to a Slack channel. It provides teams with up‑to‑date highlights without manual effort.

This workflow generates a multi-dimensional deconstruction poster for a novel. After you upload a novel file (TXT, PDF or EPUB), it extracts and cleans the text, verifies that it’s a novel, splits the content into manageable segments and uses a language model to summarise characters, events, timelines and key items. Those summaries are combined into a poster specification containing a title, character-relationship graph, a chronological timeline and additional dimensions. A Gemini image-generation tool then converts this specification into a hand-drawn style poster.

This template demonstrates how to start a workflow from both a third‑party plugin event and a custom HTTP webhook. It listens to events from an external system via a plugin (e.g., GitHub events) and to arbitrary events via a webhook endpoint. Upon receiving an event, it processes the data, optionally summarises it with an LLM, and sends a formatted notification to Slack. The example shows how to build event‑driven integrations across multiple sources.

Automatically generate production-ready Dify workflow YAML from plain text requirements. Skip the tedious drag-and-drop design — just describe your agent, and this tool outputs a fully functional, importable workflow. Supports all core nodes including LLM, Code, If-Else, Iteration, and more. The generated workflow can be further edited in Dify for iteration and optimization.

This workflow extracts structured information from images of a Chinese identity card. It processes the front (人像面) and back (国徽面) of the ID card separately, using vision‑enabled LLMs to extract fields such as Name, Gender, Ethnicity, Birth Date, Address, Issuing Authority, Validity Period and ID Number. The results are validated (e.g., checking dates and ID number format) and combined into a single JSON object. It is designed for onboarding, KYC or identity verification scenarios.

[Dify x EdgeOne]一个真正"会掷骰子"的克苏鲁神话 TRPG 守秘人。 不同于普通的 AI 角色扮演，本模板基于 COC 7e 规则做了真正的 D100 判定 —— LLM 不能撒谎说成功就成功，所有判定结果都是 Code 节点真随机生成；HP/SAN/物品栏通过 Dify 会话变量持久化，跨轮不丢；多个玩家同时玩时，每个人的角色卡按 conversation_id 完全隔离。 内置 30 分钟可通关的克苏鲁短剧本《雾港谜案》，1923 年英格兰海港背景，3 个场景 + 3 种结局。叙事节点推荐使用 DeepSeek-R1 等推理模型以获得最佳氛围。 推荐部署方式：使用 EdgeOne Pages 的 Dify Frontend Starter 模板一键部署前端（https://pages.edgeone.ai/templates/dify-frontend），全球边缘节点分发 + 多用户并发隔离。 核心特性： - COC 7e D100 真随机判定（不是 LLM 假装的成功失败） - 角色状态跨轮持久化（HP/SAN/物品栏） - 多玩家并发自动隔离 - 内置完整剧本《雾港谜案》 - 推理型 LLM 驱动的克苏鲁式恐怖叙事 - 4 分支意图分类（建角/判定/叙事/查询）

基于 Dify 与 EdgeOne 构建的 AI 剧本杀主持人应用，还原《隐秘饭店》经典剧本场景。AI 扮演 DM（主持人），负责分发角色、推进剧情、释放线索，并引导玩家进行角色扮演、推理讨论与真凶投票。支持多轮对话、剧情分支与多结局触发，适合 2-6 人在线游玩，沉浸式体验完整一局剧本杀。

[Dify x EdgeOne] Parallel Life Realm is an interactive text-based life simulator workflow. It lets users experience alternate lives through four modes: random reincarnation, custom world building, plot choices, and plot rewriting. This workflow guides users to pick a mode first, then generates a unique parallel world setting, characters, and branching storylines based on their choice.

"今天吃啥"是一个有记忆的 AI 吃饭管家。跟普通 Chatbot 不同——它不只是聊天，它通过飞书多维表格做持久化记忆存储。 每次推荐前，Agent 自动查飞书表格——你最近三天吃了什么？有什么忌口？预算偏好？然后基于这些数据生成个性化推荐。推荐完自动写回表格，下次自动避开重复食材。 你能随时打开飞书表格，看到自己的完整饮食记录——花了多少钱、吃了什么、每天营养搭配。

一个面向炉石传说玩家的卡组推荐与环境分析助手。用户可以用自然语言查询标准或狂野环境的热门卡组、职业构筑、起手留牌、对局优劣、竞技场强度、酒馆战棋流派和官方天梯排名。工作流会先判断用户需求，再按需调用对应数据接口，避免无关请求，并将结果整理成易读的中文建议。