Jump to content

Connect Leeroopedia MCP: Equip your AI agents to search best practices, build plans, verify code, diagnose failures, and look up hyperparameter defaults.

Workflow:Langgenius Dify Plugin Management

From Leeroopedia


Knowledge Sources
Domains LLMs, Plugin_System, Extensibility
Last Updated 2026-02-08 14:00 GMT

Overview

End-to-end process for discovering, installing, configuring, and managing plugins that extend the Dify platform with additional tools, model providers, agent strategies, and data sources.

Description

This workflow covers the complete plugin lifecycle in Dify. Plugins can be sourced from the official marketplace, GitHub repositories, or local package files. The installation process includes manifest validation, async task tracking, credential configuration, and permission management. Plugin categories include tools, model providers, agent strategies, data sources, and triggers. The system also supports MCP (Model Context Protocol) server integration for external tool connectivity.

Usage

Execute this workflow when you need to extend Dify's capabilities beyond the built-in features. Common use cases include adding new LLM providers, integrating external tools (search engines, databases, APIs), connecting MCP servers, installing custom agent reasoning strategies, or adding new data source connectors for knowledge base ingestion.

Execution Steps

Step 1: Plugin Discovery

Browse available plugins through the marketplace, search by category or keyword, or identify a specific plugin from a GitHub repository or local package file. Plugin categories include tool providers, model providers, agent strategies, data sources, and triggers.

Discovery channels:

  • Marketplace — Browse curated collections with category filters (Agent, Search, Productivity, RAG, etc.)
  • GitHub — Specify repository URL with version tag and package name
  • Local Package — Upload a .pkg file directly from your machine
  • Bundle — Install multiple related plugins as a single package

Step 2: Plugin Installation

Install the selected plugin through the appropriate channel. Each installation method uploads and validates the plugin manifest, then performs async installation tracked by a task system. The manifest declares the plugin's capabilities, required credentials, minimum Dify version, and resource declarations.

Installation flow:

  • Upload or fetch plugin package (produces a unique identifier and manifest)
  • Validate manifest including version compatibility with current Dify instance
  • Initiate async installation task
  • Track installation progress (uploading, readyToInstall, installing, installed)
  • Handle installation failures with error details

Step 3: Credential Configuration

Configure authentication credentials required by the plugin. Plugins may require API keys, OAuth tokens, or other authentication methods. Credentials can be set at the workspace level (shared) or per-user. Some plugins support OAuth flows with configurable client ID and secret.

Authentication types:

  • API Key (Header) — API key sent in HTTP header with configurable prefix (Basic, Bearer, or custom)
  • API Key (Query) — API key sent as URL query parameter
  • OAuth — Full OAuth flow with client configuration and workspace defaults
  • None — No authentication required

Step 4: Tool Integration

Once installed, plugins appear in the tool selector within the workflow editor and agent configuration. Tools from different collection types (built-in, custom API, workflow, MCP) can be mixed in a single application. Configure tool-specific parameters and test tool invocations.

Tool collection types:

  • Built-in — Pre-installed tools bundled with plugins
  • Custom API — User-defined tools via OpenAPI/Swagger schema
  • Workflow — Existing workflows exposed as callable tools
  • MCP — External Model Context Protocol servers providing tool access
  • Data Source — Connectors for knowledge base data ingestion

Step 5: Updates and Lifecycle Management

Manage installed plugins through updates, permission changes, and uninstallation. Configure auto-update strategies to keep plugins current. Monitor plugin health and manage workspace-level installation permissions.

Lifecycle operations:

  • Update — Upgrade from marketplace or GitHub to newer versions
  • Auto-update — Configure strategies (disabled, fix-only patches, or always latest)
  • Permissions — Control who can install or debug plugins (everyone, admins only, or nobody)
  • Uninstall — Remove plugin and clean up resources
  • Version tracking — View current vs latest version with changelogs

Execution Diagram

GitHub URL

Workflow Repository