MCP server for managing N8N workflows with HTTP streaming, OAuth 2.0 authentication, and session persistence
Server has 11 tools with basic schemas and descriptions, but significant quality gaps. All tools have descriptions (10-50 chars, below the 50-200 char baseline for LLM optimization) and input schemas are present but minimal. Parameter descriptions are sparse or missing entirely. No output schemas documented. Error handling is generic (no recovery guidance, no categorization). Tools lack annotations (readOnlyHint, destructiveHint). Naming is verb-first (good), but descriptions are too terse to guide LLM selection. No pagination support despite tools returning lists. Destructive operations (delete_workflow) lack confirmation patterns.
Activate an N8N workflow
Create a new N8N workflow
Deactivate an N8N workflow
Delete an N8N workflow
Execute an N8N workflow
Get a specific workflow execution
Get workflow execution history
Descriptions are too terse (10-50 chars vs. 50-200 char baseline). LLMs cannot distinguish when to use similar tools (e.g., activate_workflow vs. execute_workflow). Add context: 'Activate a workflow to enable scheduled/triggered execution' vs. 'Execute a workflow immediately with optional input data'.
No output schemas documented. LLMs cannot plan downstream calls or extract required fields (e.g., workflow_id from create_workflow response). Document return types for all tools.
create_workflow and update_workflow have vague parameter descriptions ('Workflow nodes', 'Node connections'). LLMs cannot construct valid payloads. Specify expected structure: 'Array of node objects with {id, type, parameters}' and 'Object mapping source node IDs to target node IDs'.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 50 | <=2025-11-25 | v2 |
Get a specific N8N workflow by ID
Get all N8N workflows
Stop a running workflow execution
Update an existing N8N workflow
Destructive operation (delete_workflow) lacks confirmation pattern. No dry-run, no confirmation step. Agents can irreversibly delete workflows without safeguards.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Clients cannot distinguish safe reads from destructive writes. Mark get_* tools as readOnly, delete_workflow as destructive, activate/deactivate as idempotent.