Talk to your servers.
Connect Claude, ChatGPT, Cursor, n8n or any MCP-compatible AI to your panel and ask in your own words: restart, check the console for an error, change a config line or back up before touching anything.
Survival is lagging. Find out why.
CPU 94%, 12× "Can't keep up!" since ClearLag was installed.
Proposal
- view-distance=16 + view-distance=10
Three steps, done
1Create a token
In your panel, under AI & copilot. Choose which servers, what it can do and for how long. Or connect Claude or ChatGPT with one click.
2Paste it in your AI
The panel gives you the configuration ready for Claude, ChatGPT, Cursor, VS Code, n8n and more.
3Ask in your own words
The AI investigates, shows you what it will change (with the diff) and does it when you say yes.
Pick your AI
The same address for all of them. The ones with OAuth connect with one click; the rest use a token you create in the panel.
Claude (web & app)
One-click connect (OAuth)- 1Nothing to create in SolumeCore first: Claude brings you to the panel.
- 2In Claude open Settings → Connectors → Add custom connector.
- 3Name: SolumeCore. URL: the MCP address below. Press Add, then Connect.
- 4Sign in to SolumeCore, choose servers and permissions, and you are back in Claude, connected.
MCP server URL
https://api.solumecore.com/mcpThings you can ask
“Why did my Minecraft server crash last night?”
“Back up Survival and then restart it.”
“Set max-players to 50 and difficulty to hard.”
“Search the log for plugin errors and tell me which one fails.”
“Install EssentialsX and LuckPerms, latest compatible version.”
“Add Steve to the whitelist and remove op from Alex.”
“How much RAM did my server use this week? Is my plan enough?”
“In Essentials config.yml, change the nickname prefix to "~".”
You stay in control
Built so an AI can help you without being able to do harm: everything it changes goes through you.
It asks first
Every change is a proposal. Approve in the chat or in the panel; destructive actions always ask.
Nothing is lost
Deleted files go to IA-deleted-files and can be restored for 3 days. Automatic backup before big changes.
Per-token permissions
Servers, areas and level (read, change, destructive) for each AI, separately.
Expires and has limits
Expiry from 1 to 365 days, daily change limit and per-minute limit.
Everything is logged
Every call with tool, parameters (no secrets), result and token.
Revoke instantly
One click and the token stops working. Optional: only from certain IPs.
The token isn’t stored
It is shown once; we only keep its fingerprint (hash).
OAuth 2.1 with PKCE
Claude and ChatGPT connect without you copying any secret.
What your AI can do
Each token only sees the ones you allow. "Read" tools never change anything.
Servers & metrics
list_serversreadget_server_statusreadget_server_metricsreadget_server_activityread
Power
start_serverchangerestart_serverchangestop_serverchangekill_serverdestructivewake_serverchange
Console & logs
read_consolereadsearch_logsreaddiagnose_crashreadsend_commandchange
Files
list_filesreadread_filereadsearch_filesreadwrite_filechangeedit_filechangecreate_folderchangerename_filechangecopy_filechangedelete_filesdestructivelist_deleted_filesreadrestore_deleted_fileschange
Backups
list_backupsreadcreate_backupchangerestore_backupdestructive
Players
list_playersreadmanage_playerchange
Plugins & mods
search_pluginsreadlist_installed_pluginsreadcheck_plugin_updatesreadinstall_pluginchangeupdate_pluginschange
server.properties
get_server_propertiesreadupdate_server_propertieschange
Scheduled tasks
list_schedulesreadrun_schedulechange
VPS
list_vpsreadget_vps_statusreadvps_powerchangevps_force_powerdestructive
Security questions
No. By default every change is first a proposal you must approve, in the chat or in the panel. Destructive actions (delete, restore a backup, force-kill) always ask, even if you turn off questions for normal changes. And every call is logged in the panel.