OpenAI has problems with the data returned by forgejo_mcp_server_version #74
Labels
No labels
Compat/Breaking
Kind/Bug
Kind/Documentation
Kind/Enhancement
Kind/Feature
Kind/OpenSpec
Kind/Security
Kind/Testing
Priority/Critical
Priority/High
Priority/Low
Priority/Medium
RFC - Request For Comments
Reviewed/Confirmed
Reviewed/Duplicate
Reviewed/Invalid
Reviewed/Won't Fix
Status/Abandoned
Status/Blocked
Status/Need More Info
hermes-attempted
hermes-needs-clarification
hermes-ready
hermes-review
hermes-wip
human-required
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
agentic-forges/forgejo-mcp#74
Loading…
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Attempting to use this MCP server with GPT5.2 (and other OpenAI models) using copilot-cli, I receive the error:
The response to the intializing
tools/listrequest does in fact return:To the best of my knowledge, JSON Schema doesn't require an unpopulated
propertiesproperty, but OpenAI seems to expect it. Adding insult to injury, the root cause of this is not in the forgejo MCP project, but in the mcp-go dependency, namely here where theProperitiesare tagged with theomitempydesignator for json serialization.hey @a2800276 thanks for bringing this up.
any proposal? do I understand correctly, that it would be nice to have the properties populated by forgejo-mcp? what shall be the content?
presumably an empty
{ ... "properties":{} ... }tbh: this is just something that came to my attention. The correct way to handle this would be to raise the issue with
mcp-goand have them remove theomitempty. I can try to build a version of forgejo-mcp with a patched mcp-go to verify and then also raise the issue upstream.There may be consequences to adding the empty
propertiesto the InputSchema that I am not aware of though...That seems like a lot of work though, and was not really what I had in mind opening this issue :) Not sure if anyone know the mcp-go maintainers? Otherwise: I'll have to see when I get around to it.
I dont know the mcp-go maintainers... I'd be happy if you contribute, but we can also keep this issue open for documentation purposes and leave the code as is. up to you ;)
Actually, this turned out to be fairly easy to set up and test. And adding an empty "properties:{}" property does seem to solve the problem. I'll post this to the gh issues and submit a PR and will link & update this issue.
(Small world incidentally, I saw you're located in Bonn, I'm in Cologne :)
Ok, I've submitted an issue and PR at the mcp-go github.
Urgh, bla.
They already fixed this 🫣
here's the fix to forgejo-mcp:
duh 🤣
I just tested this by patching the
mcp-godependency tov0.44.0and I can confirm that it's working.