I am trying to enable several workflows for the Instance-level MCP, but I can’t get them to stay in the list.
Steps I’ve taken:
I go to Settings > Instance-level MCP.
The MCP is Enabled.
I click on the orange “Enable workflows” button.
My workflows appear in the search dropdown (they are all Published and have Webhook/Schedule triggers).
I click on the workflow name in the list to select it.
The problem: After clicking or closing the search, the workflows do not appear in the main table. Only one workflow stays there, and I cannot add the rest.
Technical details:
n8n version: [Aquí pon tu versión, ej: 1.70.2]
Deployment: [Ej: n8n Cloud / Self-hosted Docker / Desktop app]
Workflows Status: They have a Description filled in their settings and are currently Published.
Triggers used: Webhook and Schedule triggers.
What I have already tried:
Refreshing the instance list using the refresh icon.
Adding a description to each workflow.
Toggling the “Published” status off and on.
What is the error message (if any)?
Please share your workflow
(Select the nodes on your canvas and use the keyboard shortcuts CMD+C/CTRL+C and CMD+V/CTRL+V to copy and paste the workflow.)
I went to an existing, publish, MCP-enabled workflow. Edited it, and had the same issue, so it is affecting existing workflows as they are modified, too, it seems.
I am having the same issue. I have multiple workflows that were ALREADY MCP enabled and were being access by claude while building (i published then bring it in to assist debug) . I updated to the newer version today and my two most recent workflows are now no longer selected as MCP enabled.
I have gone through the 3 different ways to enable MCP and it will turn on, and then turn off immediately with NO error message whatsoever.
Having the same issue on v2.8.3, self-hosted Docker on Digital Ocean with PostgreSQL.
I dug into the logs and found the cause. Every time you toggle “Available in MCP” on a workflow, n8n immediately fires Disabled MCP access for deactivated workflow in the backend logs and reverts the setting to false. This happens even when the workflow is both active (active = true in the database) and Published in the UI.
You can confirm by running docker logs <container> --tail 0 --follow while toggling the MCP setting — you’ll see that log line appear instantly.
Workaround: Force the setting directly in PostgreSQL:
sql
UPDATE workflow_entity
SET settings = jsonb_set(settings::jsonb, '{availableInMCP}', 'true')
WHERE id = '<your-workflow-id>';
You can find your workflow IDs with:
sql
SELECT id, name, active FROM workflow_entity;
After the update, the workflow appears in Settings → Instance-level MCP and is visible to connected MCP clients. The UI toggle also shows as ON.
The bug: The backend condition check for “deactivated” is incorrectly evaluating workflows that are both active and published as deactivated, causing the MCP setting to be immediately reverted on save. This seems to affect self-hosted Docker deployments — haven’t tested cloud.
