Hi Team,
I have created MCP server it was working fine previously but from yesterday its not working i tried with different MCP but i think there is some issue with UiPath side
for reference attaching screenshot
Anyone know why this happening?
Hi Team,
I have created MCP server it was working fine previously but from yesterday its not working i tried with different MCP but i think there is some issue with UiPath side
for reference attaching screenshot
Anyone know why this happening?
Try logout and login from assistant and orchestrator.
Let me know if this works
Happy Automation
Already did isuse is with npm package from UiPath side i belive
Try creating another sample MCP server and deploying. If that also don’t work, then it could be from UiPath side.
Tried that as well same issue
@loginerror could you please redirect this to correct point of contact in this case?
@akash-revansiddh.javaleka meanwhile please share all the logs you can share also the UiPath organisation name, tenant etc which will be helpful to team to troubleshoot this.
Hey @akash-revansiddh.javaleka
I will check with the team, I think it is indeed a bug.
In the meantime, one thing you might want to just give a shot (no promises), is to try to override the bad env variable via the env config of your MCP server:
"env": {
"NPM_CONFIG_IGNORE_SCRIPTS": "false", // the likely real fix
"NPM_CONFIG_MIN_RELEASE_AGE": "", // silences the invalid-config warning
"NPM_CONFIG_ALLOW_GIT": "true" // only if a dependency uses a git: spec
}
It’s a Claude Code suggestion, but I didn’t want to leave you without anything in the meantime.
Also, @akash-revansiddh.javaleka, would you mind sharing with me the /tmp/home/.npm/_logs/…debug-0.log log file? It might speed things up.
Thanks for the quick reply please find the logs from orchestrator
Error_Logs.txt (11.2 KB)
I tried this approach but still getting same error.
Let me know if you have any other workaround
Thanks
Good news, the team found and fixed the issue and it will be hot fixed.
If you are really impatient, I have another suggestion that might allow you to skip the wait time for that deployment:
Change the MCP server definition from
command: npxto a shell wrapper that points npm at a different (non-existent → empty) global config, so the broken/usr/local/etc/npmrcis never parsed:"command": "sh", "args": \["-c", "exec env NPM_CONFIG_GLOBALCONFIG=/tmp/none.npmrc npx -y @modelcontextprotocol/server-filesystem /tmp"\]
@loginerror
That’s awesome this would really helpful let me try this approach
Just a proper disclaimer: this is just a hopeful stop-gap dirty workaround that should be reverted as soon as we deploy the fix
This is not a way to run things in production.
Hi @loginerror yes understoood! Actually tomorrow we have community event on MCP session for that purpose only i was looking workaround.
Thanks & Regards,
Akash Javalekar
It worked Thank you so much for quick support