This workflow follows the Execute Workflow Trigger → HTTP Request recipe pattern — see all workflows that pair these two integrations.
The workflow JSON
Copy or download the full n8n JSON below. Paste it into a new n8n workflow, add your credentials, activate. Full import guide →
{
"id": "aicp-eval-drift",
"name": "Eval / drift lane (scheduled, AG-3)",
"meta": {
"description": "Weekly continuous-eval lane. A Schedule Trigger calls the `lanes` service (GET lanes:8081/eval-metrics?days=7, header x-lanes-token from $env.LANES_TOKEN), which runs scripts/eval_metrics.py over the audit ledger and returns analyze() as JSON. n8n cannot run python itself (hardened image \u2014 see docs/decisions/0006), so the lane runs in `lanes` while the SCHEDULE, EXECUTION LOG and Notify seam stay in n8n. A Code node formats agentic-leverage + quality metrics and any drift WARNs into { severity, title, message, source }; severity is 'warn' if any drift flag OR the ledger hash chain fails to verify OR the lane call fails. This measures metadata the router already logs; it is NOT a semantic eval \u2014 see docs/design/EVAL.md. Deploy with scripts/n8n_bootstrap.py \u2014 no UI clicks."
},
"nodes": [
{
"parameters": {
"rule": {
"interval": [
{
"field": "weeks",
"triggerAtDay": [
1
],
"triggerAtHour": 7
}
]
}
},
"id": "e1e1e1e1-0000-0000-0000-000000000001",
"name": "Weekly (Mon 07:00)",
"type": "n8n-nodes-base.scheduleTrigger",
"typeVersion": 1.2,
"position": [
220,
300
]
},
{
"parameters": {
"url": "http://lanes:8081/eval-metrics?days=7",
"sendHeaders": true,
"headerParameters": {
"parameters": [
{
"name": "x-lanes-token",
"value": "={{ $env.LANES_TOKEN }}"
}
]
},
"options": {}
},
"id": "e1e1e1e1-0000-0000-0000-000000000002",
"name": "Run eval metrics",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.2,
"position": [
440,
300
],
"onError": "continueRegularOutput"
},
{
"parameters": {
"jsCode": "// The lanes service returns eval_metrics.analyze() as the response BODY.\n// A refused ledger (chain_ok:false) or a failed lane call escalates to warn \u2014\n// the lane must never look healthy just because the call did not come back.\nconst d = $input.first().json || {};\nconst NL = String.fromCharCode(10);\nif (d.chain_ok === false || (!d.metrics && !d.note)) {\n return [{ json: {\n severity: 'warn',\n title: 'Eval lane: LEDGER REFUSED or lane call failed',\n message: ['chain_ok=' + d.chain_ok,\n d.error || JSON.stringify(d).slice(0, 400),\n 'Run scripts/verify_audit.py.'].join(NL),\n source: 'eval-drift'\n } }];\n}\nif (!d.metrics) {\n return [{ json: { severity: 'info', title: 'Eval lane: no data in window',\n message: d.note || 'no decisions in the eval window yet.', source: 'eval-drift' } }];\n}\nconst m = d.metrics;\nconst flags = d.drift_flags || [];\nconst warn = flags.length > 0;\nconst warnLines = flags.map(function (f) { return '[' + f.metric + '] ' + f.message; }).join(NL);\n// HARNESS: which CONFIGURATION produced the result, not which model. Reported\n// only when at least two configs cleared the sample floor \u2014 below that the\n// breakdown is descriptive and ranking it would be inventing a signal.\nconst h = d.harness || {};\nconst cmp = (h.comparable || []);\nlet harnessLine;\nif (cmp.length >= 2) {\n const best = (h.configs || []).filter(function (c) { return c.sufficient; })\n .sort(function (a, b) { return (b.quality_held_pct || 0) - (a.quality_held_pct || 0); })[0];\n harnessLine = 'Harness: ' + cmp.length + ' comparable configs; best held ' +\n (best ? best.quality_held_pct + '% at $' + best.cost_per_held_usd + '/held (' + best.config + ')' : 'n/a');\n} else {\n harnessLine = 'Harness: ' + (h.configs || []).length + ' config(s) seen, ' +\n 'too few runs each to compare (floor n=' + (h.min_n || '?') + ').';\n}\n\nconst lines = [\n 'Agentic leverage (proxy): ' + m.agentic_leverage_pct + '% (' + m.leverage_numerator + '/' + m.attempted + ')',\n 'Verify-pass: ' + m.verify_pass_pct + '% Change-failure: ' + m.change_failure_pct + '%',\n 'Block rate: ' + m.block_rate_pct + '% PII-block: ' + m.pii_block_pct + '% Approval: ' + m.approval_pct + '%',\n harnessLine,\n '',\n warn ? ('DRIFT (' + flags.length + '):' + NL + warnLines) : 'No drift flags (key metrics within threshold).',\n '',\n 'window ' + d.window.since + ' -> ' + d.window.until + ' n=' + d.records_in_window\n];\nreturn [{ json: {\n severity: warn ? 'warn' : 'info',\n title: 'Eval lane: leverage ' + m.agentic_leverage_pct + '%, verify-pass ' + m.verify_pass_pct + '%' + (warn ? (' - ' + flags.length + ' drift flag(s)') : ''),\n message: lines.join(NL),\n source: 'eval-drift'\n} }];"
},
"id": "e1e1e1e1-0000-0000-0000-000000000003",
"name": "Format eval summary",
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
660,
300
]
},
{
"parameters": {
"source": "database",
"workflowId": {
"__rl": true,
"value": "aicp-notify",
"mode": "id"
},
"options": {
"waitForSubWorkflow": true
},
"workflowInputs": {
"mappingMode": "autoMapInputData",
"value": {},
"matchingColumns": [],
"schema": [
{
"id": "severity",
"displayName": "severity",
"type": "string",
"required": false,
"defaultMatch": false,
"display": true,
"canBeUsedToMatch": true
},
{
"id": "title",
"displayName": "title",
"type": "string",
"required": false,
"defaultMatch": false,
"display": true,
"canBeUsedToMatch": true
},
{
"id": "message",
"displayName": "message",
"type": "string",
"required": false,
"defaultMatch": false,
"display": true,
"canBeUsedToMatch": true
},
{
"id": "source",
"displayName": "source",
"type": "string",
"required": false,
"defaultMatch": false,
"display": true,
"canBeUsedToMatch": true
}
],
"attemptToConvertTypes": false,
"convertFieldsToString": true
}
},
"id": "e1e1e1e1-0000-0000-0000-000000000004",
"name": "Notify",
"type": "n8n-nodes-base.executeWorkflow",
"typeVersion": 1.2,
"position": [
880,
300
]
},
{
"id": "ondemand",
"name": "Run on demand",
"type": "n8n-nodes-base.executeWorkflowTrigger",
"typeVersion": 1.1,
"position": [
220,
460
],
"parameters": {
"inputSource": "passthrough"
}
}
],
"connections": {
"Weekly (Mon 07:00)": {
"main": [
[
{
"node": "Run eval metrics",
"type": "main",
"index": 0
}
]
]
},
"Run eval metrics": {
"main": [
[
{
"node": "Format eval summary",
"type": "main",
"index": 0
}
]
]
},
"Format eval summary": {
"main": [
[
{
"node": "Notify",
"type": "main",
"index": 0
}
]
]
},
"Run on demand": {
"main": [
[
{
"node": "Run eval metrics",
"type": "main",
"index": 0
}
]
]
}
},
"settings": {
"errorWorkflow": "aicp-error-trigger"
},
"active": false
}
Pro
For the full experience including quality scoring and batch install features for each workflow upgrade to Pro
About this workflow
Eval / drift lane (scheduled, AG-3). Uses httpRequest, executeWorkflowTrigger. Scheduled trigger; 5 nodes.
Source: https://github.com/jgobuilds/ai-control-plane-public/blob/main/n8n-workflows/eval-drift.workflow.json — original creator credit. Request a take-down →
Related workflows
Workflows that share integrations, category, or trigger type with this one. All free to copy and import.
PF WF-02 Caption Generator v18.3. Uses microsoftSharePoint, httpRequest, executeWorkflowTrigger. Scheduled trigger; 39 nodes.
Microsoft SharePoint, HTTP Request, Execute Workflow Trigger
Deliverables → Drive (rehydrated). Uses httpRequest, executeWorkflowTrigger. Scheduled trigger; 15 nodes.
HTTP Request, Execute Workflow Trigger
Drive → workspace sync (scrubbed). Uses httpRequest, executeWorkflowTrigger. Scheduled trigger; 10 nodes.
HTTP Request, Execute Workflow Trigger
dispatch. Uses executeWorkflowTrigger, httpRequest. Scheduled trigger; 8 nodes.
Execute Workflow Trigger, HTTP Request
Daily digest lane (scheduled, human-awareness digest tier). Uses httpRequest, executeWorkflowTrigger. Scheduled trigger; 6 nodes.
HTTP Request, Execute Workflow Trigger