Week 36 / 2026.
A short record of changes between issues. Full answers and charts live on the question pages.
- Data window
- Aug 31, 2026 – Sep 6, 2026
- Last updated
- Sep 18, 2026
- Author
- Romuald Członkowski
Commentary
Between 31 August and 6 September, n8n-mcp produced 98,043 workflows, 14,006 a day on average, across 1,784,850 tool calls with a 0.26% error rate. The split interests me more than the volume: of 133,533 active installs, 9% create workflows at all, and 369 installs hold 31 workflows or more, 3% of the creators. The reach is wide; the regular building sits with a small group.
1.8% of workflows fail validation, and the smaller the workflow, the more often it fails: 3.47% at two or three nodes, 1.94% at four to ten, 0.05% at 101 to 300. The daily rate swings between 0.4% and 3%, so I don't read a single week as a trend, and the data doesn't show what drives the pattern.
The same window holds 109,604 workflow edits: 2.46 operations each on average, a median of 888 ms, a verifiable improvement in 62% of them. Changing configuration is the most common intent (42,906), adding functionality is two and a half times rarer (17,029), fixing validation accounts for 5,101. Agents spend their time tuning what already runs, not building from scratch.
In this window's corpus (39,276 workflows, median 7 nodes) Code appears in 73%, HTTP Request in 53%, an LLM model node in 9%, an AI Agent in 6%. The Data Table share has been climbing since March: 3.3% then, 5.8% in July, 9.1% in the September data. Error Trigger stays at 1.7% and hasn't moved since March (1.9% to 2.5% month to month), while a third of connection tests end in an error, half of those an address or network error.
Changes versus the previous issue
| figure | value | change |
|---|---|---|
| Workflows created | 98,043 | — |
| Workflows per day | 14,006 | -0% |
| Tool calls | 1,784,850 | -5% |
| Calls ending in error | 0.26% | +11% |
| Failed validation | 1.8% | +84% |
| Active installs | 133,533 | +11% |
| Week-1 retention | n/a | — |
| Workflows added to corpus (week) | 39,276 | +20% |
This week's answers
- Which LLM powers n8n AI agents: OpenAI, Anthropic, Gemini or Ollama?
In n8n workflows built last week, OpenAI holds 56% of model nodes, Anthropic 15%, Google Gemini 12% and OpenRouter 9%. In Mar 2026 OpenAI had 48% and Anthropic 12%.
- Which n8n nodes do AI agents use most, and which trigger nodes start their workflows?
The most used node in AI-built n8n workflows is Code, present in 73% of last week's workflows, followed by HTTP Request (53%) and Webhook (47%). The median workflow has 7 nodes.
- Which n8n node is growing fastest?
The fastest-growing n8n node is Data Table: its share of AI-built workflows rose from 4.3% in May 2026 to 8.6% in Aug 2026, a 2.0× increase. Convert to File is second (1.8×). Last week Data Table was the most searched node among AI agents.
- Do AI-built n8n workflows handle errors?
Rarely. 1.7% of AI-built n8n workflows from last week contain an Error Trigger node, and 2.0% of those built in Aug 2026. Since March the share has never exceeded 2.5%.
- Do AI agents build n8n workflows from nodes or from code?
From code. The Code node appears in 73% of AI-built n8n workflows from last week and HTTP Request in 53%. In Aug 2026 the Code share was 73%. Agents would rather write JavaScript and call an API than look for a ready-made integration.
- How large are AI-built n8n workflows, and which fail validation?
49% of workflows created last week have at least 11 nodes and 19% have 31 or more. The larger the workflow, the less often it fails validation: 3.5% for 2–3 nodes versus 0.27% for 31–100.
- Which n8n nodes do AI agents search for?
Last week the most searched n8n node was data table (297 installs), ahead of webhook (180) and schedule trigger (100). The ranking counts unique installs, not queries.
- How do AI agents (Claude Code, Cursor) build n8n workflows through MCP: build or maintain?
Maintain. Of 1,784,850 tool calls last week, 60% read workflows, executions and lists, and 15% write. Fetching executions alone is 31% of calls. In the same week 98,043 workflows were created, 14,006 a day.
Most searched nodes
data table (297) · Webhook (180) · schedule trigger (100) · google sheets (82) · http request (79) · Gmail (77) · Telegram (68) · openai (54) · ai agent (51) · form trigger (49)
$env expressions blocked in Code nodes
$env is not available inside Code nodes for security reasons. Access environment variables through the n8n UI or pass them via workflow-level static data. Use $input or hardcode values if needed.