Skip to content
MCP server

TypeSafe

Calibrated yes/no, choice and score answers from TypeSafe's Jev, for 1 to 1000 rows

Runs locallyStatus and history
Install
Pricing
Paid
Platforms
CLI
Runs
Locally
Version
0.1.3
Listing updated
2 days ago

About TypeSafe

TypeSafe is a paid MCP server for ai models and llm tooling. It runs locally on your machine and is started by your AI client.

TypeSafe's Jev model answers typed questions with a calibrated confidence: how likely a yes/no statement is to hold, which of your options fits, or where something sits on a scale you define. This server passes those questions through from an assistant, and its batch tool takes up to 1000 rows at once so postings, transactions or emails can be triaged without the assistant reading each one.

Every tool is a read, nothing changes anywhere, and none is flagged as money. Calls are billed by input tokens, which the page puts at a fraction of a cent each, and Jev answers in about half a second. The batch tool can write its results as JSONL to an outbox directory and reads input files from the paths allowed by MCP_FILE_ROOTS.

You need a TypeSafe account and an API key from its console, stored in a file. Without a key the server still starts and lists its tools, and each call answers with a configuration error naming the variable. A health check shows the API in use, the default model and the models your key reaches.

Install

Run in a terminal

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

Add to claude_desktop_config.json

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

Add to ~/.cursor/mcp.json

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

Add to .vscode/mcp.json

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

Add to ~/.codeium/windsurf/mcp_config.json

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

Add to your client's MCP settings (most use this shape)

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

Configuration

  • TYPESAFE_API_KEY_FILEPath of a file holding your TypeSafe API key.Required

Tools this server provides (1)

Taken from the project documentation; the running server may differ.

ask_batch
Puts the same questions to up to 1000 rows from an items file, optionally saving results as JSONL.

What it does

Calibrated yes/no answers
Returns how likely a statement is to hold.
Multiple-choice routing
Picks which of your options fits an item.
Custom scales
Places something on a scale you define.
Batch of 1000 rows
Runs the same questions over up to 1000 rows in one call.
JSONL output
Saves batch results to a file in the outbox directory.
Health check
Shows the API, default model and models your key can reach.

Good for

  • List remote senior roles from a file of 300 job postings
  • Decide whether a support email is urgent and which team should take it
  • Categorise a month of card transactions
  • Rank reviews by likelihood the customer will leave
  • Flag passages that answer a question, with the uncertain ones marked for review

Pricing

Paid

Billed by input tokens; prices not published on the page

Prices change. Check foreverworks.github.io before you buy.

TypeSafe FAQ

How is it billed?

By input tokens, which the page describes as a fraction of a cent each.

Which model is used by default?

jev-latest, which tracks the newest Jev; it can be changed with TYPESAFE_MODEL.

Can a reseller's key be used?

No. Only a key from TypeSafe's own console works, unless the call goes through the reseller's gateway.

Does it change any data?

No. Every tool is a read.

Is TypeSafe free?

No. TypeSafe is a paid product.

How do I add TypeSafe to Claude Code?

Run this command in your terminal: claude mcp add typesafe --env TYPESAFE_API_KEY_FILE='<TYPESAFE_API_KEY_FILE>' -- npx -y @foreverworks/mcp-typesafe

What are the best alternatives to TypeSafe?

The closest alternatives to TypeSafe are Claude Code Ultimate Guide, signafAIde, Toolkien, Lugh and HypePaper.

Alternatives to TypeSafe

  1. MCP server for searching the Claude Code Ultimate Guide from any MCP client

    Runs locally
    • Open source
    • 6.5K GitHub stars
    • AI models and LLM tooling
  2. signafAIde

    MCP server

    Local MCP server to set up, verify and check coverage of signed agent action records

    Runs locally
    • 5 tools
    • Freemium
    • Security
  3. Toolkien

    MCP server

    One MCP gateway for all your MCP servers, with code mode, permissions and audit

    Not checked yet
    • 3 tools
    • Remote
    • Open source
    • AI models and LLM tooling
  4. Lugh

    MCP server

    Write, check and dry-run Lugh agent programs with policy, budgets and approvals

    Runs locally
    • 5 tools
    • Open source
    • AI models and LLM tooling
  5. HypePaper

    MCP server

    Research papers, benchmarks, citations and topic experts for AI assistants

    • 58 tools
    • Remote
    • Freemium
    • AI models and LLM tooling