Hey everyone!
I built an MCP server that lets you test, score, and inspect n8n workflows directly from Claude (or any MCP-compatible client).
What it does
- Config-driven test suites — define payloads in JSON, run them via webhook or execute endpoint
- Automatic scoring — two-tier system: Tier 1 checks infrastructure (HTTP, timeout, output), Tier 3 checks quality (field validation, contains, min_length, etc.)
- Execution traces — lightweight per-node summary without loading full execution JSON
- Node catalog — search 436+ nodes, validate types, get suggestions for natural-language tasks
- Safe by design — no credential management, no auto-fix loops, no destructive operations
Example test config
{
"workflowName": "my-webhook",
"triggerMode": "webhook",
"webhookPath": "/webhook/my-handler",
"qualityThreshold": 85,
"testPayloads": [
{ "name": "happy-path", "data": { "message": "Hello" } },
{ "name": "empty-input", "data": { "message": "" } }
],
"tier3Checks": [
{ "name": "has-response", "field": "output", "check": "not_empty", "severity": "error" }
]
}
Scoring output
{
"passed": true,
"score": 93,
"tier1Score": 100,
"tier3Score": 80,
"issues": [{ "tier": "tier3", "severity": "warning", "check": "response-length", "message": "Response too short" }]
}
Quick setup
git clone https://github.com/Souzix76/n8n-workflow-tester-safe.git
cd n8n-workflow-tester-safe
npm install && npm run build
node dist/index.js # MCP server (stdio)
Works as MCP server for Claude Code or as standalone CLI for CI pipelines. 19 tools total — testing, workflow operations, introspection, and catalog helpers.
MIT licensed. Feedback welcome!