otherapi_key

Dify

Run and inspect a published Dify Workflow app through its Service API, including configuration and workflow run history.

Verdict

Dify is a low-code AI workflow builder that lets teams chain LLM calls, API requests, and conditional logic into reusable apps. This MCP exposes four tools: inspect a workflow's configuration, trigger a run, check its status, and list past executions. Marketing and product teams use it to automate content generation pipelines or data enrichment flows without writing code. One caveat: runs execute blocking mode only, so long workflows will hold your thread until they finish or fail.

Common use cases

  • Generate blog outlines from research notes
  • Enrich lead records with company data
  • Translate product copy into five languages
  • Score customer feedback sentiment at scale
  • Chain web scraping to summary generation

Integration

Vendor
Dify
Category
other
Auth
API_KEY
Tools
4
Composio slug
dify

Tools

  • Get Workflow App Configuration

    Return the connected Workflow app's identity, declared input form and upload limits, tool icons, and hosted-site settings. If the app is unpublished, the response identifies the unavailable parameters section while preserving the other meta

  • Get Workflow Run

    Return one workflow run's status, inputs, arbitrary outputs, error, and execution metrics by run ID.

  • List Workflow Runs

    Return one page of run summaries for the connected Workflow app, with filters and a continuation cursor for older or additional runs.

  • Run Workflow

    Run the connected app's latest published workflow in blocking mode and return its app-defined outputs. Inspect the app configuration first: the configured workflow may perform irreversible external side effects.

Setup

Setup guide

  1. 11. In Switchy, open Settings → Integrations → Add MCP Server and select Dify. 2. Log into your Dify workspace and navigate to the workflow app you want to connect. 3. Open the app's API settings and generate a new API key with workflow execution permissions. 4. Paste the key into Switchy's configuration dialog and save. 5. Inside any Project, type '@Dify get workflow app configuration' to confirm the connection returns your app's input schema and upload limits. 6. If the response shows your workflow's declared inputs, the integration is live and ready to trigger runs.

What teammates see: by default, memories from Dify are scoped to the Project (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.

Works well with

Top models

Compatibility data appears once enough Projects have used this MCP together with a given model.

How Switchy teams use it

Not enough Projects yet to publish anonymised usage stats (we require ≥ 50 Projects per week).

Starter prompts

Inspect Workflow Schema

@Dify get workflow app configuration so I can see what inputs this workflow expects and what file types it accepts.

Run Content Pipeline

@Dify run workflow with the input topic set to 'remote team collaboration tools' and return the generated article text.

Check Run Status

@Dify get workflow run for run ID abc123 and tell me whether it completed successfully or hit an error.

List Recent Executions

@Dify list workflow runs for the past week and show me the input parameters and execution time for each.

Automate Data Enrichment

@Dify run workflow with the domain input set to 'acme.com' and return the enriched company profile as JSON.

Example outputs

Illustrative - representative of the model's voice and quality, not literal recordings.

Prompt

@dify get the configuration for our customer-onboarding workflow and tell me what inputs it expects

Output

Your customer-onboarding workflow app expects three required inputs: 'customer_email' (string), 'plan_tier' (enum: starter|pro|enterprise), and 'signup_date' (ISO 8601 timestamp). Optional inputs include 'referral_code' and 'custom_fields' (JSON object). The workflow is published and configured to upload files up to 10MB. It exposes two tools in its execution: a Salesforce contact creator and a welcome-email sender via SendGrid.

Notes

This read-only query reveals the workflow's input contract and integrated tools before you execute it. Essential for understanding what data the AI needs to gather before calling 'Run Workflow'. If the app is unpublished, the configuration response will flag unavailable parameters—you cannot run draft workflows via this MCP.

Use-case deep-dives

Post-deployment workflow smoke test

Running a live workflow test after a Dify update

A 5-person engineering team just updated a Dify workflow that generates PDF invoices from Stripe webhooks. Before merging the feature branch, the tech lead asks Switchy to run the workflow with a test payload and return the output. The MCP's Run Workflow tool executes the latest published version in blocking mode and surfaces the PDF URL in under 10 seconds. The team confirms the invoice renders correctly and ships. This scenario has a hard constraint: the workflow must be published, and the run is not a dry-run—it hits live external services. If your workflow sends emails or charges cards, test with a sandbox Stripe account first. Best for teams that treat Dify workflows as black-box services and need a quick integration test without opening the Dify console.

Frequently asked

What does the Dify MCP let me do in Switchy?

It connects your Dify Workflow apps so AI agents in Switchy can trigger runs, check execution status, and retrieve outputs. You configure the workflow in Dify's studio, then Switchy calls it like a function — useful for chaining LLM tasks, data transforms, or external API calls that Dify orchestrates.

Do I need a Dify API key to connect this MCP?

Yes. Generate an API key in your Dify workspace settings, then paste it into Switchy's connection form. The key scopes to a single Workflow app, so if you want agents to call multiple workflows, connect one MCP instance per app or rotate the key as needed.

Can the MCP modify or delete my Dify workflows?

No. It's read-only for configuration and read-write for runs. Agents can start new executions and poll results, but they can't edit the workflow graph, change variables, or unpublish versions. You still own the workflow logic in Dify's studio.

Why use this instead of calling Dify's API directly?

The MCP wraps pagination, error parsing, and blocking-mode polling so your prompts stay simple. Instead of writing retry loops or cursor logic, you tell the agent "run the sentiment workflow on this text" and it handles the HTTP dance. Useful when the workflow is stable but the inputs vary.

Who on my team should connect the Dify MCP?

Whoever owns the Dify workspace and can generate API keys. Once connected in Switchy, any team member with access to the shared workspace can invoke the workflow through agents — they don't need their own Dify login or key.

More other MCPs

See all other MCPs →

Data last verified 61 hours ago.Sources aggregated hourly to weekly. See docs/architecture/model-directory.md.