# TypeSafe MCP server

> Lets an AI client obtain calibrated yes/no judgments, multiple-choice selections, or numerical scores for single items or batches of up to 1000 rows.

- Repository: https://github.com/foreverworks/mcps
- Page: https://hostedmcp.org/servers/foreverworks/typesafe
- Transport: stdio
- Stars: 5
- License: MIT
- Latest version: 0.1.3

## Install

### Claude Code

```sh
claude mcp add --env 'TYPESAFE_API_KEY_FILE=<TYPESAFE_API_KEY_FILE>' typesafe -- npx -y @foreverworks/mcp-typesafe
```

### Cursor (`~/.cursor/mcp.json`)

```json
{
  "mcpServers": {
    "typesafe": {
      "command": "npx",
      "args": [
        "-y",
        "@foreverworks/mcp-typesafe"
      ],
      "env": {
        "TYPESAFE_API_KEY_FILE": "<TYPESAFE_API_KEY_FILE>"
      }
    }
  }
}
```

### VS Code (`.vscode/mcp.json`)

```json
{
  "servers": {
    "typesafe": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@foreverworks/mcp-typesafe"
      ],
      "env": {
        "TYPESAFE_API_KEY_FILE": "<TYPESAFE_API_KEY_FILE>"
      }
    }
  }
}
```

### VS Code CLI

```sh
code --add-mcp '{"name":"typesafe","type":"stdio","command":"npx","args":["-y","@foreverworks/mcp-typesafe"],"env":{"TYPESAFE_API_KEY_FILE":"<TYPESAFE_API_KEY_FILE>"}}'
```

### Claude Desktop (`claude_desktop_config.json`)

```json
{
  "mcpServers": {
    "typesafe": {
      "command": "npx",
      "args": [
        "-y",
        "@foreverworks/mcp-typesafe"
      ],
      "env": {
        "TYPESAFE_API_KEY_FILE": "<TYPESAFE_API_KEY_FILE>"
      }
    }
  }
}
```

### Codex

```sh
codex mcp add typesafe --env 'TYPESAFE_API_KEY_FILE=<TYPESAFE_API_KEY_FILE>' -- npx -y @foreverworks/mcp-typesafe
```

### Codex config.toml (`~/.codex/config.toml`)

```toml
[mcp_servers.typesafe]
command = "npx"
args = ["-y", "@foreverworks/mcp-typesafe"]

[mcp_servers.typesafe.env]
TYPESAFE_API_KEY_FILE = "<TYPESAFE_API_KEY_FILE>"
```

### Gemini CLI

```sh
gemini mcp add -s user -e 'TYPESAFE_API_KEY_FILE=<TYPESAFE_API_KEY_FILE>' typesafe npx -- -y @foreverworks/mcp-typesafe
```

### Gemini settings.json (`~/.gemini/settings.json`)

```json
{
  "mcpServers": {
    "typesafe": {
      "command": "npx",
      "args": [
        "-y",
        "@foreverworks/mcp-typesafe"
      ],
      "env": {
        "TYPESAFE_API_KEY_FILE": "<TYPESAFE_API_KEY_FILE>"
      }
    }
  }
}
```

## Environment variables

| Name | Required | Secret | Description |
|---|---|---|---|
| `TYPESAFE_API_KEY_FILE` | yes | no | Path of a file holding your TypeSafe API key. |

## Tools (3)

- `health`
- `ask`
- `ask_batch`

## FAQ

### What kind of answers can I get from TypeSafe?

You can get 'noul' (probability of a yes/no statement), 'choice' (one of 2 to 255 options), or 'score' (a value on a 2 to 10 level scale).

### How many rows can I process at once?

You can process up to 1000 rows in a single batch request.

### Does using TypeSafe change my data?

No, all tools provided by the server are read-only operations.
