XIO

xWire

Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.

n8n Community - Latest topics

Latest topics

Agents (preview): how to deliver an n8n Agent from a dev instance to a prod instance? "Couldn't load agent details" after moving the workflow

Describe the problem/error/question We run two self-hosted n8n instances on the Business plan: n8n-dev (development) and n8n-prod (production). On n8n-dev we built a first-class n8n Agent (preview) called Main — Rokodo - ISO27001 - Agent. It is called from a workflow: Chat Hub → Guardrails - Input → Agent node (calls the n8n Agent) → Reply - Agent / Reply - Blocked / Reply - Unavailable When we move this workflow to n8n-prod, the workflow itself arrives, but the Agent node shows "Couldn't load…

Creator Portal False Rejection for @creative-x-lab/n8n-nodes-17track

I submitted my community node package @creative-x-lab/n8n-nodes-17track (latest: 1.1.5) for verification through the Creator Portal. The automated pre-check keeps rejecting it with “Missing credential test”, but the scan command the Portal recommends passes all checks on every scanner version I can find: npx @n8n/scan-community-package@beta @creative-x-lab/n8n-nodes-17track Provenance check passed for “@creative-x-lab/n8n-nodes-17track@1.1.5” Fetched source from…

Postgres connections with custom SSL certs finally work properly (3.1)

Most of my client instances run on managed Postgres (RDS and a couple of self-hosted boxes), and until recently the Postgres credentials in n8n only handled the simple cases. If the database required a custom CA or client certs, the workarounds were ugly: relax security, or route everything through an SSH tunnel and hope. n8n 3.1 (Nov 3) fixed this properly. Postgres credentials now take a CA certificate, a client certificate, and a client private key. The fields show up when SSL mode is set to…

Help - all my workflows are down

Describe the problem/error/question None of my workflows are triggering. The AI said to have the owner restart the instance, which I did and it didn’t help. I can’t get anyone in support to actually support their product. No cap has been reached. Manual trigger works. What is the error message (if any)? Please share your workflow No errors. They just aren't triggering. Share the output returned by the last node Information on your n8n setup n8n version: Database (default: SQLite): n8n…

N8n Forms + Nextcloud WebDAV: a German working-time checker (ETag locking, Form node gotchas on 2.46.6)

I built a small n8n workflow that records working hours and absences through two n8n Forms and checks them against the German Working Hours Act (ArbZG). I’m sharing it mainly for the things that cost me time on n8n 2.46.6. What it does Two Form Triggers (hours, absences) in one workflow, five triggers in total with the Friday report and the Monday demo reset. Saves everything to a zeiten.csv in Nextcloud over WebDAV. Checks the rules in plain code (no LLM): max 10 h/day (avg 8 h) per §3, breaks…

Why my approval webhook kept leaving people with a spinning browser (Sharing my learnings)

Hey n8n Community, Been building a multi-step approval system, one-click approve and reject links in emails, tokens, the usual. Three things caught me out. Email security scanners pre-fetch your links. Corporate mail filters follow URLs to check them before the recipient ever sees the message. If your approve link is a single-use token, the scanner burns it. Approvers also double-click buttons. Either way the second request finds the token already consumed. Which then hangs the browser. When my…

Routing Carrier Tracking Webhooks Into One Shipment Status Feed in n8n: Where Custom Logistics Software Development Fits

A logistics company may work with several carriers, each sending shipment updates in a different format. One carrier reports a package as IN_TRANSIT, another uses MOVING, and a third sends separate events for departure and arrival at a sorting facility. When these updates are stored in separate systems, dispatchers and customer support teams struggle to get one consistent shipment view. n8n can receive carrier webhooks, normalize their payloads, and route the events into a shared shipment…

Would you let an AI agent fix your production n8n workflows without your approval?

AI agents are getting pretty good at finding errors, rewriting logic, and suggesting fixes. But here’s where things get interesting. Imagine an AI agent detects a problem in your n8n workflow. It identifies the broken node, modifies the logic, runs a test, and everything passes. Then it automatically publishes the updated workflow. Sounds great, right? Until you realize the fix changed something else. Maybe the CRM now writes to a different field. Maybe duplicate records get created. Or maybe…

Different behaviour for HTTP GET nodes when running workflow vs running node individually

Very long title, but an absolutely bizarre bug I’m grappling with which I haven’t been able to figure out in ~3 days now. I have an online service with no API (e.g. https://foo.bar), I have used a Puppeteer node to do auth flow, then capture cookies, samlCookies, csrfTokens etc. - all that fun stuff. This auth flow has been working flawlessly for the past few months now with no hiccups. I am now trying to set up a new GET node to read a page on the same service (https://foo.bar/singlePage).…

Would you let an AI agent modify your production n8n workflows without approval?

Your production workflow breaks. An AI agent reads the error, finds the cause, and patches the workflow. The next execution succeeds. Everything looks fixed. Except the patch also removed the filter that prevented duplicate customer messages. The workflow runs again. It just doesn’t behave correctly anymore. AI agents that can build and modify n8n workflows are getting good. But who checks the agent that’s fixing your automation? My position is, AI proposes, it doesn’t deploy. AI proposes…

[Sharing] Multi-Platform Lead Capture workflow (Webhook + Conversation Memory + Google Sheets)

Hi everyone, I’m sharing a Multi-Platform Lead Capture workflow I built. What it does It collects leads through a conversational flow using a webhook. It works with any platform that can send a POST request (WhatsApp, Telegram, website forms, etc.). It collects: Name Email Phone (optional – user can type “skip”) What they need Source It also has conversation memory using Google Sheets, so it remembers previous answers. Features Multi-stage conversation (welcome → name → email → phone → need)…

If Node Problem-Emails nicht sortiert

Hallo zusammen, ich habe ein Problem mit meinem E-Mail-Sortierungs-Workflow in n8n. Ich habe einen IF-Node eingerichtet, der E-Mails mit dem Eintrag Severity: info erkennen und diese automatisch aus dem Posteingang in einen bestimmten Ordner verschieben soll. Allerdings funktioniert die Filterung nicht wie gewünscht. Der IF-Node unterscheidet offenbar nicht zwischen den verschiedenen Severity-Leveln, sodass aktuell alle E-Mails in denselben Ordner verschoben werden. Hat jemand eine Idee, woran…

Nod Questions - managing context window size?

Nod along with the new AI agent builder has bought me back to n8n! So much has changed, great work everyone! I have installed Nod and it’s working great. I stated this thread as I see no other threads around Nod and the discord sever also has no Nod dedicated thread, at least that I could find. My 1st question, does one need to manage context window size with Nod? I can find no slash command or button to clear a session or start a new one? Is this because its not needed or is there some other…

Zendesk is retiring API tokens. Your n8n workflows keep running until they don't

If you touch Zendesk on client instances, this one is worth ten minutes this week. Newer n8n versions now default new Zendesk credentials to OAuth2, and Zendesk is retiring the API token method on their side. Existing token credentials keep working for now, so nothing breaks today. The failure comes later and it comes quietly: the token dies or a brand new token simply cannot be created the old way, and a scheduled workflow starts skipping runs before anyone notices. What I do on each instance…

How Do You Handle Client Credentials When Building n8n Automations for Non-Technical Clients?

I’m building an AI Automation where I develop and deploy workflows using my own self-hosted n8n instance rather than asking every client to maintain their own n8n infrastructure. One challenge I’m trying to solve is client credential management and onboarding. Most of my target clients are non-technical business owners (dentists, real estate agencies, clinics, small businesses, etc.). For example, imagine I build an automation that connects: Google Sheets Gmail Google Calendar Google Drive…

1. Faceless YouTube Shorts in n8n: a CSV of topics in, a Short per row in Google Drive, with Claude and Shotstack (free n8n workflow template)

Hey n8n community, if you run a faceless facts or explainer channel, here’s a free n8n workflow template for faceless YouTube Shorts: a CSV of topics in Google Drive in, one finished vertical Short per row out, saved back to a Drive folder. Claude Opus 5.5 writes the script and the whole edit, and Shotstack films a scene for every sentence, voices it, adds word-by-word captions and renders the video, ready for YouTube Shorts, Instagram Reels and TikTok. It’s a rebuild of a popular gallery…

Não está sendo gerado output

o nó é executado com sucesso mas <No fields - item(s) exist, but they’re empty!-- n8n Cloud users: If you’re experiencing issues with your cloud instance (e.g., can’t sign in, instance unavailable), please contact support at help@n8n.io → What is the error message (if any)? Please share your workflow No fields - item(s) exist, but they're empty Share the output returned by the last node Information on your n8n setup n8n version: Database (default: SQLite): n8n EXECUTIONS_PROCESS setting…

N8N Agent Slack Chat response fix?

Hi, I’m using an n8n Agent with the Slack channel for an HR assistant. In one-to-one DMs, the agent always posts its reply in a thread under the user’s message. We’d like it to reply directly in the DM instead. Workaround tried: a Slack tool (send message to user) so the agent posts the answer in the DM, with instructions to leave the final response empty. The DM message works, but the channel still posts something in the thread (the agent’s plan, “Done”, a ✓, or a short greeting in…

Automatizar

Estimados expertos. Estoy en el proceso del aprendizaje con n8n. Y quisiera validar la posibilidad de crear un flujo que me permita automatizar la compra de productos en un supermercado. Me estoy imaginando lo siguiente: Un hoja en google sheet que contenga [“Nombre del producto”, “Cantidad”, “MarcadelProducto”] Una URL a la pagina del supermercado. Agregar los productos que estan en la hoja de googlesheet y agregarlos al carrito de compras. Me gustaria tener una idea de como podria hacer eso y…

Weekly UK public tenders digest: open tenders matching your keywords, emailed every Monday (importable JSON)

I built a workflow for anyone who bids for UK public-sector contracts and is tired of checking Find a Tender and Contracts Finder separately. Every Monday it emails one table of open tenders published in the last 7 days that match your keywords. What it does Schedule trigger (Mondays 8am) plus a manual trigger for testing. A Set node holds your keywords (comma-separated), a minimum contract value and the email address. An Apify node runs UK Government Tenders Scraper, which searches both…

Reasoning_effort error with n8n Assistant and OpenAI-compatible endpoint

Describe the problem/error/question Dear fellow n8n users, I testing n8n Assistant in a self hosted n8n instance with an OpenAI-compatible endpoint (Langdock) and Daytona. The setup seems complete without any errors but if I try to send a message to Assistant, I’m reveicing this error: Error: 400 Function tools with reasoning_effort are not supported for gpt-6.1-terra in /v1/chat/completions. To use function tools, use /v1/responses or set reasoning_effort to ‘none’. I’ve tried different models…

Here's a dangerous habit in automation that nobody talks about enough. 😂

Your workflow passes 500 tests. One day, a test fails. The output changed. So you update the expected result, rerun the test, and everything goes green again. Congratulations. You might have just taught your testing system to accept a bug. Here’s the uncomfortable part: Who actually decided the new result was correct? The developer? The client? The AI? Or did everyone just assume the latest successful execution was the new truth? An automation can run perfectly while producing completely wrong…

Loop node issue: It outputs more than one item when it is done

I’ve had this issue with the Loop node for a long time and wanted to ask about it. I can’t recall when it started, but the Loop node now outputs more than one item when it is done. I remember the Loop node always output one item when it was done in the old version. To fix this issue, I tried adding a Limit node after the loop and limiting the output to only one item, specifically the first item. This is because I have a Google Sheets Get Row action after it. However, it doesn’t really help,…

How Are You Preserving Context When Handing Work Between AI Agents?

I’m interested in how people manage context when an AI coding workflow involves multiple agents. For example: Agent A investigates a bug. Agent B implements the fix. Agent C reviews the changes. A developer takes over if something fails. The difficult part isn’t necessarily triggering the next agent. It’s making sure the next step receives the right information. A useful handoff record could contain: Original task and expected result. Investigation findings. Files changed and implementation…

N8n Cloud Form Trigger – Submission Fails Despite Authentication Set to None

Title: n8n Cloud Form Trigger fails with “Problem submitting response” despite Authentication = None Hello n8n team, I’m experiencing a persistent issue with the Form Trigger in my n8n Cloud workspace. Workflow: Fekra Tube - Unified Content Production (Manual Request) - v8 Quran Source Attribution Fix Configuration: Form Trigger authentication: None Respond When: Form Is Submitted Form Path: fekra-tube-idea Using the Test URL (/form-test/) What happens: I click Execute step on the Form Trigger.…

N8n AI Agent + Gemini OpenAI-compatible API: 400 Bad Request when executing Gmail tools

n8n AI Agent returns 400 Bad Request when using Gmail tools with Gemini through OpenAI-compatible API I’m using n8n 2.47.5 (self-hosted) with an AI Agent connected to an OpenAI Chat Model node configured to use Google’s Gemini OpenAI-compatible API. My configuration: Base URL: https://generativelanguage.googleapis.com/v1beta/openai/ Model: gemini-3.5-flash-lite Use Responses API: OFF Gemini API key configured in n8n. The issue: The model works correctly for normal text conversations through my…

Self-conflicting webhook path after Unpublish: cannot republish same workflow (self-hosted CE)

Describe the problem/error/question A previously published production workflow cannot be republished after being unpublished. The Publish validation says its POST webhook path conflicts with the same workflow (the conflict dialog links back to that workflow), not another workflow. We are seeking a supported, non-destructive recovery procedure, not a workaround requiring container recreation or direct database writes. Error Conflicting Webhook Path for POST m-workflows/book-slot. The conflicting…