How can I use the Human Input node when calling the Dify workflow via the MCP server?

[cislove98]

Self Checks

• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:)
• Please do not modify this template :) and fill in all the required fields.

1. Is this request related to a challenge you're experiencing? Tell me about your story.

I have activated the MCP server in the workflow and intend to execute it from LangGraph's ToolNode.

At this point, the workflow is called normally and enters a pending state at the human input node. However, how can I actually verify the form content configured at the human input node and proceed to execute it at the next node?

Currently, the delivery method for the human input node is set to web app.

2. Additional context or comments

workflow config

MCP Server

Workflow Log

[dosubot[bot]]

When your workflow reaches the human input node via MCP server and enters a "pending" state, here's how to interact with it:

1. Detect the Pending State

When using streaming mode, you'll receive a human_input_required event containing the form_id, form_content, inputs, actions, and expiration_time. You'll also get a workflow_paused event with status: "paused".

2. Retrieve Form Content

Use the pause details endpoint:

GET /console/api/workflow/{workflow_run_id}/pause-details

This returns form details including form_id and a backstage_input_url.

3. Submit Human Input

To submit the form and resume the workflow:

POST /api/form/human_input/{form_token}

With body:

{
  "inputs": { "field_name": "field_value" },
  "action": "Approve"
}

The form_token is sufficient for authorization—no API key is required. The workflow automatically resumes after submission.

4. Get Final Output

Since workflow resumption is asynchronous via Celery, poll the status endpoint:

GET /api/workflows/run/{workflow_run_id}

Wait until status changes from running to succeeded, then read the outputs field.

There's also a related discussion on detecting Human Input Node states that covers similar ground.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[cislove98]

POST /mcp/server/{MCP Server code}/mcp
When a workflow is executed via the MCP server, events are not received. How can I detect the pending state?

[dosubot[bot]]

현재 MCP 서버 엔드포인트는 워크플로우를 동기적으로 실행하고 최종 결과만 반환하도록 설계되어 있습니다. Human input node에서 일시정지 상태를 감지하거나 human_input_required/workflow_paused 이벤트를 수신하는 기능은 MCP 엔드포인트에서 지원되지 않습니다.

Human input node가 포함된 워크플로우의 경우, MCP 서버 대신 표준 Workflow API를 사용하는 것을 권장합니다:

1. POST /v1/workflows/run으로 워크플로우 실행
2. GET /apps/{app_id}/workflow-runs/{run_id}/pause-info로 일시정지 상태와 form_token 조회
3. POST /api/form/human_input/{form_token}으로 human input 제출
4. GET /v1/workflows/run/{workflow_run_id}를 폴링하여 완료 상태 확인

자세한 내용은 Human Input Node 감지 관련 토론을 참고하세요.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}