Python script custom return types
Hi, I am not sure if this is the intended but I am unable to return custom data types from my Python scripts that windmill is executing. Is this intentional or am I missing a step?
For example, I am trying to return a pandas data frame and it is actually returning a json array. Also when trying to return a sqlalchemy engine it just returns a string as if you tried to print the engine. Thank you...
HTTP-Route & preprocessor
I'm trying to built a little API proxy (filter the content before the client gets it) using windmill and stumbled upon issues/questions i can't find the answers to:
- Do HTTP routes somehow support wildcards/catch-alls? Would come in handy, since my little API proxy needs to handle requests to different endpoints but needs to be transparent for the client aka act like the client is talking to the upstream API. Otherweise i would have to create multiple HTTP triggers, which is fine, but not as convenient.
- If i set the http route to Authentication No Auth, is the Bearer token not forwarded to the preprocessor, so i can evaluate it there?...
App: Required input fields
What is the recommended way of validating input fields when a button is clicked and before the underlying script/flow is called?
For example, I have an app with:
- several text inputs
- a button that triggers a script using the values from the text input fields...
How to collect details of steps of for loop.
Given the flow of the for loop. The collection of the iter will result in an array of undefined because the unpublish step returns undefined. However I want to have more control over the collection of the results, I want the details of productSet mutation in this case. The publish/unpubish do not give results they can only fail.
How would one easily built something like that. Having to create a seperate function to merge stuff seems very, uhm empty
//collector.ts...
Nuget version not found issue
Is there any cache for nuget sources in csharp?
Version 1.0.7 exists in nuget source, but windmill is unable to find this version, and sometimes it mentions the Nearest version available is 1.0.5, sometimes it mentions 1.0.6, seems very unstable.
Sometimes after a few minutes, windmill is able to find the version 1.0.7 ...
...
Error Ducklake: Writing parquet file fails.
ExecutionErr: execution error:
"HTTP Error: Unable to connect to URL http://localhost:36867/api/w/testsr2/s3_proxy/_default_/datalake/main/states2/ducklake-01997aa2-1513-75ee-a66e-ff5a6e3d9aa3.parquet?uploads=: Forbidden (HTTP code 403)"
Metadata within Postgres (Neon) looks good, S3 Browser up- and download looks good. Using a stand alone DuckDB / Ducklake with same S3 and Postgres infrastructure works just fine, so I think, something is wrong with my Windmill configuration.
...
Error: Connecting to database: pool timed out while waiting for an open connection
Windmill server and worker containers deployed to ECS exit with the above error. The RDS postgres logs show
could not receive data from client: Connection reset by peer, suggesting a connection is being made but the client (server/worker containers) are killing the connection.
RDS instance is db.t4g.large. Confirmed the max connections setting is still the default (which allows ~900 connections for this instance size).
ECS tasks are deployed on Fargate with 2 vCPU and 4GB Memory. I don't see any evidence of memory or CPU constraints in monitoring. ...Docker
I'm having difficulty figuring out how do handle a very simple scenario.
I have a flow where a step generates a file, and pushed it to S3.
I have a subsequent step where I need to run a docker image that would run a script to download that file and execute docker code on it. Since I cannot mount any files, how can I download a file and have it available to the code inside the docker container. What is the general process for executing a docker image that needs to have access to a file?...
Parsing resource content $res
Hi there, I am using windmill v1.513.1 and I define an Input which has type array and item type "Items are resources". Now, the ressources are passed to the next step as something like "$res:/path/to/resource". This $res is breaking everything. Is it intended to appear and how do I parse the resource in a subsequent script using the go-library's GetResource(string) ? Thanks for hints
Using perpetual script to trigger flow
I want to create a system that will listen to an event from an external service and trigger a flow to process the data. My initial through is to use a perpetual script as the listening service and trigger the flow when an event was captured from the listening service. However, I can't seem to find a way to trigger a flow using code.
what is the best way to setup a listening service that can utilize flow? Is there a way to natively trigger flow using something like inner flow mechanism but for script?...
Using custom components in an app
Anyone have any experience with making custom components? I am wondering what is common practice when making exchanging info between the component and a script? Currently i am setting the output of the component to my "question" then a background script sees a change in the output and runs with that new output from the component. In the script that branches out to all those actions thats needed. Is there some easier way?
I will give an example so its easier to undestand. Lets say i want a component to look up adresses. By default the component is in its search state, then by writing into a searchfield it will set an output and a background script will run with those perameters, eg {"action": "search_query", "query": "abc"}. the script will then do some things and will return something like {"action": "search_suggestion", "suggestions": [x, y, z]}, this will then trigger a recalculation of the component and it will see those perameters and rerender itself. This will go on in a "loop", back and forth where everything is stateless.
I am wondering if there is an easier approach? Do people use APIs? In that case how, and can you use background scripts in an app for that?...
Find jobs completed with script_hash returns SQL error
I don't know am I using the API somehow wrong, but if I try to list completed jobs by script hash, I get following error.
For example, curl -X GET --url "https://app.windmill.dev/api/w/${WM_WORKSPACE}/jobs/completed/list?script_hash=80c5f49a98fb00f0
Return response:...
MCP version update?
The current implementation of MCP at Windmill works on SSE. However, newer MCP clients have forced Streamable HTTP, preventing connections.
Information: https://modelcontextprotocol.io/specification/2025-06-18/basic/transports#streamable-http...
Can't select scripts to add to flow
I have written a bunch of new graphql scripts and when I try to create a new flow and add the script it does not show up in the library or when I search. I have been using windmill for a while now and this is the first time that my files are not showing up file making flows so I think it might be a bug
Script Cache
In testing caching for script results, I noticed my script that is tagged to run on a Windows Agent executes the script each time it is run. The same script without a tag (I.e. using regular workers) does return the cached result immediately.
Is this expected behavior? My goal was to cache output from a script run against our active directory to populate a dashboard. It doesn't need to be executed with every app refresh because the data changes infrequently....
I was an admin of our Windmill account and now I don't even see it. Was our account deleted?
We just paid for 3 users on Sep 6th and I was one of the Admins, yet now it looks like our account doesn't exist or I've been kicked out of it.
Color change AgGrid Table
Hello everyone, is it possible to change the background color of a specific row in an AgGrid Table?
Agent Tags
Based on the docs one should be able to restrict workers to specific workspaces using custom tags like “tag(workspace1)” as an example. I deployed a windows agent with a tag “ad-dev(infra)” but users in other workspaces are able to use that tag in their scripts. Please advise if this is not the correct path for me in trying to restrict which workspaces are allowed to use the windows agents I have configured.
Input doesn't get preprocessor output in the flow
I have a flow that is triggered by the email trigger.
I've added a Preprocessor step that maps the trigger event to the other data structure. Let's say, preprocessor returns this object:
{ account: "Account Name", amount: "5" }
When I test the preprocessor, it works as expected. ...
