dify

What This Skill Does

The dify skill serves as the primary operational interface for managing self-hosted Dify instances. It leverages the dify-manager MCP server to bridge the gap between human intent and Dify's management and console APIs. This skill is engineered to handle high-level application orchestration, dataset management, and knowledge-base maintenance. It is specifically designed to separate operational tasks—such as updating a prompt or checking dataset health—from the more complex, structural task of authoring Workflow DSLs. By providing a structured framework for interacting with your deployment, it ensures that your self-hosted environment remains stable, documented, and efficiently managed without requiring manual intervention in the backend.

Installation

To integrate this skill into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/arn0ld87/dify-orchestrator Ensure your local environment has the dify-manager MCP server configured at /Users/alexanderschneider/mcp-servers/dify-manager to enable full functionality.

Use Cases

• Lifecycle Management: Create, list, or delete Dify applications directly through the agent.
• Content Engineering: Update and version control system prompts for chat-based or completion-style applications.
• Knowledge Management: Automate the creation of datasets, trigger document uploads, and verify the indexing status of your files.
• Operational Audits: Perform health checks on your self-hosted instance and verify if specific features from new Dify releases are compatible with your current deployment version.
• Retrieval Optimization: Query existing knowledge bases to validate if the RAG system is returning the expected context for specific user queries.

Example Prompts

1. "Check the current health of our self-hosted Dify instance and list all active applications."
2. "Update the system prompt for the 'Customer Support' application to include the new privacy guidelines."
3. "Is the 'Multi-Step Query' feature supported in our current Dify version, and can you link the 'Product Manual' dataset to our main bot?"

Tips & Limitations

Always prioritize official documentation and current Dify release notes before performing configuration changes. This skill is strictly for management operations. If you find yourself needing to edit YAML or JSON structures for workflows, nodes, or edges, delegate that task to the dify-workflow skill. Never expose API keys or sensitive environment variables in your prompts or log outputs. When in doubt regarding a destructive action, like deleting an application or dataset, the agent will provide a clear impact analysis before proceeding.