A few weeks ago n8n released the ability to create and update workflows using the official n8n MCP server, and I know many of you are using it already.
We’d love your feedback and thoughts so far (I’m from n8n):
General feedback
How reliable has workflow creation been for you? Are the generated workflows landing close to what you’d build manually, or are you spending a lot of time fixing things up after? Any specific pain points or things that surprised you?
Tips & tricks
If you’ve found prompting patterns, workflow structures, or setups that consistently get better results out of the MCP server, share them here. Would love to start collecting what’s actually working.
Skills
If you could add anything to an official n8n MCP server skill, what would it be? Think node coverage, error handling, testing, documentation generation , or whatever you’ve been wishing it could do. And if you’ve already built your own custom skills, drop them below - I would love to see what the community is doing with it.
Good, bad, or weird, all feedback is appreciated!
Thanks!
Ophir
p.s.
If you provide any feedback, Please add the version number you’re using - thanks!!
6 Likes
Overall, workflow generation has been quite impressive. For straightforward automations, it gets very close to what I’d build manually.
However, for more complex flows (branching logic, error handling, or multi-step integrations), I still end up tweaking quite a bit after generation.
A few things I noticed:
- Node configurations are usually correct, but edge cases (like pagination or retries) are often missing
- Credentials handling sometimes needs manual correction
- Complex IF / Switch logic can become slightly messy in generated workflows
- Still, it saves a lot of initial setup time, which is a big win
Tips & Tricks
What worked better for me:
- Being very explicit about input/output structure for each step
- Breaking down requests into smaller workflows instead of one big prompt
- Mentioning error handling requirements upfront (like retries, fallback paths) improves results a lot
- Asking for “production-ready workflow with logging + error handling” gives noticeably better output
Skills Suggestions
Would love to see:
- Built-in error handling templates (retry patterns, dead-letter flows, etc.)
- Better testing / dry-run mode for generated workflows
- Automatic workflow documentation generation (very useful for teams)
- Stronger support for complex branching logic + state management
Also, a “validation step” before creation (like checking for missing credentials or invalid node configs) would be super helpful.
1 Like
Been using it for a while and yes its a great thing to work with it and already made the work a lot easier in my agency but we still have to change something to get the best out of the workflows especially the error triger nodes etc.