n8n-cli-deps

name	n8n-cli-deps
description	Map n8n workflow dependencies as a graph. Shows sub-workflow calls, credential refs, and webhook URLs as tree, mermaid, or JSON. Use when understanding interconnections before making changes.
user_invocable	true

/n8n-cli-deps — Workflow Dependency Graph

Build and visualize the dependency graph across all workflows in an n8n instance.

Procedure

1. Fetch all workflows:

   • n8n-cli --json workflows list
   • For each workflow, n8n-cli --json wf get <id> to get full node data

2. Extract dependencies for each workflow:

   • Sub-workflow calls: nodes with type containing executeWorkflow — record the workflowId parameter
   • Credential refs: any node with a credentials.<type>.id field — record the credential ID
   • Webhook paths: webhook trigger nodes — record the path
   • HTTP Request URLs: outbound HTTP calls — record the URL host (the third-party services this workflow depends on)
   • Schedule triggers: cron expression (so we can later cross-reference with /n8n-cli-schedule-audit)

3. Build the graph in memory:

   • Nodes: workflows, credentials, webhook paths, external hosts
   • Edges: workflow → sub-workflow, workflow → credential, workflow → webhook, workflow → external_host

4. Render in the format the user asked for (default: tree). Options:

Tree format (default)

[Workflow A] (active)
├── calls → [Workflow B]
├── calls → [Workflow C]
├── uses → [Stripe API credential]
├── uses → [Slack OAuth credential]
├── exposes → POST /webhook/foo
└── calls out → api.stripe.com, hooks.slack.com

[Workflow B] (inactive)
└── ...

Mermaid format

graph LR
  WF_A[Workflow A] -->|sub-workflow| WF_B[Workflow B]
  WF_A -->|sub-workflow| WF_C[Workflow C]
  WF_A -->|credential| CRED_STRIPE[Stripe API]
  WF_A -->|webhook| HOOK_FOO[POST /webhook/foo]

JSON format

{
  "workflows": [...],
  "credentials": [...],
  "webhooks": [...],
  "edges": [
    {"from": "wf_id_a", "to": "wf_id_b", "type": "sub_workflow"},
    {"from": "wf_id_a", "to": "cred_stripe", "type": "credential"},
    ...
  ]
}

Output format

• Tree is best for readability, default for human use
• Mermaid is best for sharing with the team (pastes into Notion, GitHub, etc.)
• JSON is best for piping into another tool

If the user asks for a "map" or "diagram", default to mermaid. If they ask "show me what depends on X", do a single-workflow tree rooted at X.

Tips

• For instances with 100+ workflows, the full tree is huge. Offer to filter:
  • "Show me deps for workflows tagged 'production'"
  • "Show me deps for workflow X and everything it touches"
  • "Show me orphan workflows" (workflows with no incoming or outgoing dependencies — likely standalone)
• The most useful question this skill answers is: "if I delete this credential / change this webhook / deactivate this workflow, what breaks?" — that's the /n8n-cli-impact skill, which uses this graph as input.
• Sub-workflow calls can be deeply nested. Limit recursion to depth 5 to avoid huge trees.
• Save the graph to ~/.cache/n8n-cli/deps.json so /n8n-cli-impact can reuse it without re-fetching.