n8n 2.42.3: an agent platform focused on self-hosted stability
Release 2.42.3 of n8n, a visual builder for AI agents and workflows, is out. Six releases have accumulated since the last review: 2.41.3, 2.41.4, 2.41.5, 2.41.6, 2.41.7 and the current 2.42.3. The focus in them is on fixes in the core and API rather than new features.
What's new
- The task runner survives unhandled promise rejections — the core no longer stops because of a random error in asynchronous code, which matters for long-running agent processes.
- Global members can grant variable permissions for API keys — access control for sensitive data has been extended for API integration.
- Fixed execution return with incomplete trace context — fewer lost results when the observability layer fails.
- Successful database ping counting is now more resilient to event loop delays — fewer false alarms about the DB being unavailable under load.
- Queue job results are stored only for the processes that queued them — cleaner state when scaling horizontally.
What the platform does
n8n is a fair-code platform for building and deploying AI agents and workflows. A visual canvas is combined with the ability to write your own code in JavaScript and Python and to connect npm packages. Multi-step agent scenarios with logic, tool calls, human approval and full observability are claimed. The platform supports self-hosted and cloud deployment, role-based access and auditing.
Your own model instead of the cloud
You can connect your own model via the N8N_INSTANCE_AI_MODEL_URL environment variable — it accepts the base URL of any OpenAI-compatible endpoint, including LM Studio. When the variable is set, requests to the model go to that URL instead of the built-in provider. It is installed via Docker with a single command or an installation script. License — Sustainable Use License: internal use, non-commercial distribution and personal purposes are allowed; distribution is allowed only free of charge and non-commercially.
How it differs from its neighbors
Among Dify, Sim and FastGPT, only n8n states human approval — human approvals in multi-step agent scenarios — in its documentation. At the same time, custom tools and functions, a knowledge base (RAG) and sandboxed code execution are not mentioned in n8n's documentation, whereas most of its neighbors have them. Graphs, a visual editor and self-hosted via Docker are claimed by all four platforms.
| n8n | Dify | Sim | FastGPT | |
|---|---|---|---|---|
| Type | builder | builder | builder | builder |
| Language | TypeScript | TypeScript | TypeScript | TypeScript |
| License | Sustainable Use Licen… | Apache 2.0 with add. terms… | Apache-2.0 | FastGPT Open Source L… |
| GitHub stars | 206,788 | 156,267 | 29,663 | 29,690 |
| Releases in 90 days | 58 | 5 | 85 | 20 |
| Custom tools and functions | — | yes | yes | — |
| MCP support | — | yes | — | — |
| Agent memory | — | — | yes | — |
| Knowledge base (RAG) | — | yes | yes | — |
| Sandboxed code execution | — | yes | yes | — |
| Human approval | yes | — | — | — |
| Graphs and state persistence | yes | yes | yes | yes |
| Visual editor | yes | yes | yes | yes |
| Tracing and debugging | yes | yes | yes | — |
| Voice and speech | — | yes | — | — |
| Own server and Docker | yes | yes | yes | yes |
| Images and documents | — | — | — | yes |
| Own model: Ollama | — | — | yes | yes |
| Own model: vLLM | — | — | yes | — |
| Own model: LM Studio | yes | — | — | — |
| Own model: Xinference | — | — | — | yes |
| Own model: OpenAI-compat. API | yes | yes | — | yes |
Cross-checked against the projects' READMEs and documentation as of 10/07/2026: "yes" means the capability is stated there, a dash means it is not mentioned (which does not mean the capability is absent). Stars and releases — according to GitHub data.
Updating is worthwhile for those who run n8n on their own hardware and have encountered task runner stops or lost results when tracing fails. If you are only just looking at agent builders and human approval in the chain matters to you, n8n is worth exploring. You can follow new releases of platforms for running on your own hardware in the CPU3D news feed.
