When you're running n8n workflows, Claude Code sessions, Pydantic agents, RAG pipelines, and MCP servers across multiple environments, you don't have a command center. You have a dozen browser tabs, scattered config files, and no single view of what's actually failing. AgeniusDesk is the dashboard that sits on top of all of it. One interface. Every workflow, every agent, every execution, every error, across every platform you run. You see the full picture without switching contexts or losing your place. What It Solves The moment something breaks at 2am, you don't want to log into three different tools, hunt through executions, and manually cross-reference which step failed. AgeniusDesk pulls your errors automatically, surfaces them with full context, and lets you navigate straight to the source, one click. Fix it, reload it, move on. It also solves the sprawl problem. MCP servers, API keys, model providers, secrets, agent frameworks, knowledge sources: all the connective tissue that makes your stack work, scattered across config files and env vars. AgeniusDesk puts a management layer over all of it. Standout Features - Multi-instance dashboard — all your n8n instances, agent sessions, and pipelines in one view. Live execution health, workflow counts, error totals. No tab-switching. - Error management with drill-through — errors sync automatically across platforms. Click any error, navigate directly to the source. Bulk-clear by workflow. Purge in one action. - Agent visibility layer — monitor Claude Code sessions, Pydantic AI agents, and code-based agent pipelines from a single pane. See what's running, what's stalled, and what's throwing errors. - Framework compatible — built to manage most agentic platforms, not just n8n. Plug in the frameworks you're already using without rearchitecting your stack. - Knowledge source management — connect, monitor, and manage your knowledge bases and RAG sources across agents and workflows. - Code Lab — write, edit, and inject JavaScript into live n8n Code nodes directly from the dashboard. No back-and-forth. - MCP server management — connect any data source. Add servers, manage tokens, see available tools per server. - Model agnostic — configure any AI provider per instance. OpenAI, Anthropic, Ollama, whatever comes next. Switch without touching env files. - Vault-level secrets — full Infisical integration alongside the local store. Generate, rotate, and manage API keys from one UI. - Open source — self-hosted, your data stays yours.

Coming soon! It's in private beta now. Stay tuned :)

Real talk: how many people here are building automations for actual enterprise clients? Not solopreneurs or small businesses, but companies with 50+ employees, compliance requirements, multiple departments, the whole thing. If you are, what does your stack look like? How are you handling security reviews, change management, and the "we need IT to approve this" conversations? I ask because I manage operations across a portfolio of 6 MSPs and build automations daily. n8n handles the repetitive scheduled stuff and structures data for the other agents downstream, but a lot of what I run is pure Python agents. Once you're past a handful of clients and dealing with real enterprise requirements, the organization and management problem gets serious fast. Been building tooling around this and curious if anyone else is hitting the same wall.

Quick question for the agency builders here: what's your setup for managing n8n across multiple clients? I've been going back and forth between a few approaches (dedicated instances, shared instance with naming conventions, Docker per client) and every one has tradeoffs. Curious what's actually working for people running 5+ clients.

I recently got into 3D printing and started an Instagram account for it. Problem is, I'm COO for a portfolio of 7 MSPs. I don't have time to post consistently, so I automated the entire thing. The system: Self-hosted n8n workflows handle the full content lifecycle. Two scheduled triggers fire daily (10am + 3pm ET), cycling through 5 content branches: Tips, Models, Product Design, Tech, and Art. Each branch hits a specialized AI prompt that returns a caption, hashtags, image prompt, and voiceover script as structured JSON. All branches merge into one shared publish pipeline. Adding a new content type means one new node, zero changes downstream. KIE.ai generates images, a polling loop monitors status, then it publishes to Facebook and Instagram via the Graph API and logs everything to Airtable. I'm also running GPT-4o-mini vs Claude Haiku side-by-side as an A/B test. Video pipeline: Runway generates 5 to 10-second vertical clips. Suno generates music. ElevenLabs handles voiceover. Remotion composites it all with branded overlays and lower-thirds, then publishes as Reels and Facebook Videos automatically. Real costs per post: - AI caption (GPT/Claude): $0.002 - Image gen (KIE.ai Nano Banana 2, 2K): $0.06 - Video gen (Runway 5s 720p): $0.06 - Music gen (Suno): $0.06 - Voiceover (ElevenLabs): $0.006 - Image post total: ~$0.06 - Video post total: ~$0.13 14 image posts/week = $0.87. Add videos and I'm under $2/week. Stack: n8n · GPT-4o-mini · Claude Haiku · KIE.ai · Runway · Suno · ElevenLabs · Remotion · Meta Graph API · Airtable · Telegram · Proxmox Check it out in action: https://www.instagram.com/agenius3d. Only the very first post was manual. Everything else you see was created and published by this workflow. Happy to break down any part of the build. Drop a comment and I'll dig into whatever you want to see.