Skip to content

llms-install.md

Latest commit

 

History

History
33 lines (22 loc) · 1.37 KB

llms-install.md

File metadata and controls

33 lines (22 loc) · 1.37 KB
Raw

Installing the Fractera MCP server (for Cline)

Fractera is a remote / hosted MCP server. There is nothing to install or build locally and no Docker image — you only add its URL to Cline's MCP settings.

One-step setup

Add this entry to Cline's MCP settings file (cline_mcp_settings.json):

{
  "mcpServers": {
    "fractera": {
      "type": "streamableHttp",
      "url": "https://www.fractera.ai/api/mcp"
    }
  }
}

That is the entire installation. No API key, token, or environment variable is required — the connector is open.

Verify

After saving, Cline should show the fractera server as connected and list its tools: register_and_deploy, retry_deploy, check_status, get_vps_recommendation, get_subdomain, get_project_info.

A quick read-only check: ask Cline to call get_project_info — it returns the project's architecture and FAQ and needs no credentials.

What it does

Fractera deploys a private AI coding workspace (5 AI engines + an autonomous Hermes agent + private LightRAG memory) onto the user's own Ubuntu VPS in about 10 minutes, straight from chat. The deploy is IP-first: it finishes on plain HTTP at http://<your-ip>:3002. register_and_deploy is destructive by design (it wipes the target server before installing). Read-only tools need no credentials.

Knowledge base & security FAQ: https://www.fractera.ai/mcp-info