Docs MCP server
{
"mcpServers": {
"cerebrium-docs": {
"type": "http",
"url": "https://cerebrium.ai/docs/mcp"
}
}
}{
"mcpServers": {
"cerebrium-docs": {
"type": "http",
"url": "https://cerebrium.ai/docs/mcp"
}
}
}{
"mcpServers": {
"cerebrium-docs": {
"type": "http",
"url": "https://cerebrium.ai/docs/mcp"
}
}
}{
"servers": {
"cerebrium-docs": {
"type": "http",
"url": "https://cerebrium.ai/docs/mcp"
}
}
}{
"mcpServers": {
"cerebrium-docs": {
"type": "http",
"url": "https://cerebrium.ai/docs/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 0.1.1
What it does
MCP server that provides live search over Cerebrium's documentation, covering deployment, cerebrium.toml configuration, hardware options, and endpoints. Intended for developers using AI coding agents to build and deploy workloads on the Cerebrium serverless GPU/CPU platform. Also accepts feedback submissions.
- Search Cerebrium documentation to find the correct cerebrium.toml keys, defaults, and accepted values for a deployment configuration.
- Look up available GPU compute identifiers, regional availability, and per-GPU limits when selecting hardware for a workload.
- Retrieve troubleshooting guidance for failed builds, cold-start issues, or queued requests on Cerebrium.
Tools this server exposes
| Tool | Does |
|---|---|
| search | Searches the hosted Cerebrium documentation and returns matching content. |
| feedback | Sends feedback about the documentation or MCP server. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.