Skip to content
mcp/harbor
VS Code

evalpair for VS Code

io.github.srikarmk/evalpair

Run LLM eval suites on local or hosted models; compare runs with intervals and a noise floor.

client:VS Code transport:stdio runtime:pypi

Install evalpair in VS Code

.vscode/mcp.json 

{
  "inputs": [
    {
      "type": "promptString",
      "id": "evalpair-home",
      "description": "EVALPAIR_HOME",
      "password": true
    },
    {
      "type": "promptString",
      "id": "evalpair-api-key",
      "description": "EVALPAIR_API_KEY",
      "password": true
    }
  ],
  "servers": {
    "evalpair": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "evalpair"
      ],
      "env": {
        "EVALPAIR_HOME": "${input:evalpair-home}",
        "EVALPAIR_API_KEY": "${input:evalpair-api-key}"
      }
    }
  }
}

VS Code prompts for each input the first time the server starts, so no secret is written to the file. VS Code docs

evalpair in other clients

evalpair MCP server