#general

Jan 25January 26, 2026Jan 27Latest
KK
Karthik Karthik1:21 AMOpen in Slack
Hello everyone, I have an update to share.
What was failing
During local startup, I kept hitting ERRCONNECT errors because the application couldn’t reach required services when I used tilt up.
I am on Windows.
How I debugged and fixed it
• I used kubectl port-forward to expose the services to localhost.
• I ran the backend and frontend separately.
Current workflow
Terminal 1: Kubernetes (apply, verify, and port-forwards)
Terminal 2: Backend dev server
Terminal 3: Frontend dev server
Result: The backend connects cleanly (no ERRCONNECT), the frontend loads correctly, and I can now iterate locally while continuing to rely on Kubernetes for infrastructure.
🙌1
2 replies
S
Hey all, how's everyone doing?
1 reply
RG
👋 Hi everyone!
1 reply
DD
Debojyoti De Majumder6:29 AMOpen in Slack
Hi all
1 reply
?
do we need to make our own mcp server for the wemakedevs hackathon or use mcp servers from archestra online catalog
2 replies
anyone reply
please
V
Is there any hackathon going on?
VA
veer adyani12:48 PMOpen in Slack
Hey folks, quick heads up while working on this issue: https://github.com/archestra-ai/archestra/issues/2410
I ran into a Windows clone/checkout failure with error: invalid path because one of the files under app/app/mcp-catalog/data/mcp-evaluations/ had a ?tab=... in the filename, which Windows does not support.
As a temporary workaround, I renamed the file in my fork so I could clone the repo and continue working.
I also checked if this rename would impact anything in the codebase. I ran git grep "mobb-dev__bugsy" and the only matches are inside the JSON files themselves. Nothing else in the repo references these filenames, so renaming should be safe and should not break any code.
There are currently two related JSON files in that folder, and both are standalone entries, so the rename is just to make cloning work smoothly for Windows developers.
If you are okay with it, I can open a small PR to rename the file in the main repo as well. 😃
Thanks!
2 replies
DD
Dhyaneesh DS1:30 PMOpen in Slack
Hey, found Archestra through the hackathon, crazy work.
Quick question, should hackathon projects focus on end-user apps or MCP middleware [https://gofastmcp.com/servers/middleware] ?
1 reply
MK
Matvey Kukuy (archestra team)2:52 PMOpen in Slack
Welcome to all new community members! How is archestra working for you? We're actively looking for bugs to fix 🙂
🙌1
1 reply
VA
veer adyani10:35 PMOpen in Slack
Hey Archestra team,
I’m working on this issue
The Terraform provider currently relies on the following fields and methods, which now cause compile errors because they’re missing from the latest OpenAPI spec and client:
profileTool.AllowUsageWhenUntrustedDataIsPresent (field missing)
profileTool.ToolResultTreatment (field missing)
d.client.GetPromptWithResponse (method missing)
d.client.GetPromptsWithResponse (method missing)
d.client.GetPromptVersionsWithResponse (method missing)
client.CreateChatApiKeyJSONRequestBody expects *string, but provider uses string
IsOrganizationDefault (unknown field in request/response, now missing)
r.client.SetChatApiKeyDefaultWithResponse (method missing)
These have been removed from the OpenAPI spec and backend, so the provider no longer compiles against the latest API. Do you think we should proceed with refactoring the Terraform repo to align with the new API shapes, or is there another plan? Let me know your thoughts.
Here is the backend PR link for this issue (Ready for review) : https://github.com/archestra-ai/archestra/pull/2396
1 reply

Read-only live mirror of Archestra.AI Slack

👋Join the discussion withAI enthusiasts!