Joly0

Hey, i have seen you recently added a global mcp server configuration. One thing though, i have seen http type mcp servers have basically no additional things to configure, but http (and i think this applies to sse servers aswell) can have authentication (mainly through headers), which currently wouldnt work with the way those are configured. For example in my claude.json file i have an mcp server configured like this: "mcpServers": { "Coding": { "type": "http", "url": "https://mymcp.mydomain.org/Coding/mcp", "headers": { "X-API-Key": "RANDOMAPIKEY" } } },It would be great, if this was possible to configure (you could keep that simple and just allow headers in general to be configurable and the user has to handle which and how these headers are added, but thats up to you).

Reserved

Reserved

Reserved

If you are able, please report issues or feature requests via GitHub; support will still be provided here but GitHub is the preferred method. https://github.com/Joly0/unraid-pangolin_cli/issues Pangolin CLI An Unraid plugin that installs the Pangolin CLI and adds a settings page to connect your server to a Pangolin instance as a machine client. This lets you reach private resources directly on your Unraid host. From Pangolin themselves: https://docs.pangolin.net/manage/clients/understanding-clients Installation Community Application Install via the Community Applications plugin in unRAID by going to the 'Apps' tab and searching for 'Pangolin CLI'. Manual Install Install manually by navigating to: Plugins → Install Plugin and entering the plugin URL: https://github.com/Joly0/unraid-pangolin_cli/raw/main/pangolin_cli.plg

Hey guys, i noticed after rebooting my unraid server, HA-in-a-box is not able to start the vm and throws this error: Version check passed. Continuing... Host path for '/domains' is /mnt/user/domains Highest Q35 machine type available is pc-q35-10.2 The default VM network type is eth0 Putting default ha vm icon in place Icon copied successfully. Home Assistant directories are already present...continuing. There is already a vdisk image here...skipping. As this is an OVMF VM, I need to create an NVRAM file. Creating now .... Formatting '/etc/libvirt/qemu/nvram/d8e05b6b-07e0-4a41-b1b4-2a961f908de2_VARS-pure-efi.fd', fmt=raw size=65536 All required files are present. Attempting to define the VM... Failed to define the VM. There was an error during the VM definition process. Not sure though how to solve this problem, i have no idea why its failing there and how to fix it. Starting the container manually works though and starts the vm properly

Hey guys, someone maybe knows why these "yunkan" containers keep re-appearing under "recently added" every few days? The templates are found here GitHubyunkan-unraid-templates/templates at main · mrtian2016/yu...云瞰 YunKan — Unraid Community Applications 模板仓库(CPU / OpenVINO / CUDA 3 个 amd64 变体) - mrtian2016/yunkan-unraid-templatesbut i cannot see any reason why they are re-added all the time, i just see they update the "changes" part in the templates when its updated. Its super annoying to see them every few days again and again.

Just looked through it, couldn't find it, then realized I linked the wrong repository (but still correct for container templates). The actually file you should edit is this one: https://github.com/Squidly271/AppFeed/blob/3127c70c026d256115bce4d64f3777afdd6c0862/statistics.json#L1856

The repository is here https://github.com/Squidly271/Community-Applications-Moderators If you search for your username in the repo, you will find the blacklist file I was mentioning. Just delete the part for your main branch and PR it

For your information, squid has the community applications moderations repository on his GitHub, which includes everything for CA ,including blacklisted stuff. You basically have to create a PR to un-blacklist (remove your repository from the blacklist file) and let squid merge it. It's super simple and straightforward. I have done that myself 3 times already (not for plugins, but for docker container templates, which got falsely marked as duplicates)

Hey, i noticed after a reboot the agent is simply gone. When i try to access it for example through the normal unraid webui terminal and run something like "claude" i get this error: bash: /usr/local/emhttp/plugins/unraid-aicliagents/agents/claude-code/node_modules/.bin/claude: No such file or directory The "agent store" says its installed. I either have to uninstall and reinstall or (if available) update to get claude (and others) back working.

Hey ,after the latest update something broke regarding removing and creating zram swap. So i had one running before the update. Then updated and wanted to re-create the swap file to use the higher swappiness (as far as i could read in the changelog, this is only applied to new zram´s?). I removed the existing zram, which looked to be ok, then the page had to be reloaded, but after reload it still showed the "remove" button. After doing that for a another 3-4 times, it finally reloaded and showed the "create" button instead of "remove", but when i clicked create, the log showed that it couldnt create it, because the file already exists. I tried thisn a few times over an over, always getting the same problems with removing and re-creating and finally gave up. Then randomaly i looked on the dashboard card and saw, that the zram appears to be created correctly with the correct 150 swappiness.

Yes, trying to from the default unraid shell, would be great if there was a way to do this for testing and debugging (currently trying to troubleshoot working with an claude code to opencode bridge). Another thing: This has been running for over an hour, it was stuck for most of the time now at the "Backing up active RAM sessions..." step. Not sure how long it will take for the next step, but this rough for a plugin update

Also another thing i have noticed: When i have to debug mcp servers using npm/npx they do not use the correct paths and usually defualt back to /root/ directory instead of the /tmp/unraid-aicliagents/work/root/. Even when starting them like this /usr/local/emhttp/plugins/unraid-aicliagents/bin/npx it doesnt work either and they still usually place their place under /root/. It would be great here if there was some way to easily get into the same environment as cli´s would run under. I assume you are doing some path shenanigans or something similar

Btw, just did the latest update with the migration and everything and afterwards the directory /tmp/unraid-aicliagents/work/root/ was empty, so all sessions, configurations, etc were gone. I had to manually copy back those files from the backup location under /boot/config/plugins/unraid-aicliagents/persistence/root