I have given my AI agent access to the Microsoft SQL node as a tool. However, when the agent executes poor SQL, the whole workflow stops because the Microsoft SQL tool generates an error. It should be possible for the agent to see this error and correct itself but it cannot because the Microsoft SQL node does not have a “Continue on fail” option.
.
I would really appreciate the work of the person who can add this functionality!
The idea is:
An AI agent that can correct poor SQL
My use case:
Build valid SQL and not have the whole workflow stop if there’s an error
I think it would be beneficial to add this because:
Everyone can use AI agents to generate and self-repair poor Microsoft SQL queries
Any resources to support this?
Non required
Are you willing to work on this?
If you have any questions by all means let me know. If I had the skills to work on it, I would be coding it in right away!
This is a key need, especially with the rise of AI agents generating SQL dynamically. Being able to continue despite errors would allow the agent to self-correct and prevent the entire workflow from stopping. Adding a “Continue on fail” option to the Microsoft SQL node would be a real boost to make the system more robust and flexible.
That’s a solid workaround. Wrapping the SQL execution in a separate workflow also means you keep full control of what happens when the query fails, instead of letting the whole run crash.
If you can, could you share your example (workflow JSON export, or at least a screenshot of the nodes and key settings)? This “SQL tool wrapper” pattern will definitely be useful for others building AI Agents.
