Skip to content

Nerf Watch

by tokenusage.dev · remote · no sign-in to list tools · v1.0.0 · published by dev.tokenusage

Daily Nerf Watch scores for Claude and GPT models on one private, code-graded suite. Read-only.

Read only
CompareRegistry entry

2 tools: 2 only read.

Scanned Oct 21, 2026

No check matched anything in the tool definitions. No finding is not a safety guarantee. How the checks work.

claude mcp add --transport http nerfwatch https://tokenusage.dev/mcp/nerfwatch

Tools

Open a tool for its reasons and full description. Descriptions are the publisher's own words, shown as they are.

2 tools

Only reads 2Claude Code: allow
  • get_nerfwatch_historyRetained public Nerf Watch score history for one model, oldest first, including each point's suite version.No findings

    Why it is in this group

    • The name "get_nerfwatch_history" says "get": a word for reading.
    Description, as the server wrote it
    Retained public Nerf Watch score history for one model, oldest first, including each point's suite version. The required model id is available from get_nerfwatch_scores. Missing days are missing, not zero; comparisons are within one suite version. A lower score is a lower score on this suite, not proof that a provider changed a model. Unofficial, not affiliated with OpenAI or Anthropic. Source: https://tokenusage.dev/nerfwatch

    Annotations the server declares

    • title: Nerf Watch history
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • get_nerfwatch_scoresLatest public Nerf Watch scores, previous runs, changes and baselines for at most 100 models, ordered by provider, name and model id.No findings

    Why it is in this group

    • The name "get_nerfwatch_scores" says "get": a word for reading.
    Description, as the server wrote it
    Latest public Nerf Watch scores, previous runs, changes and baselines for at most 100 models, ordered by provider, name and model id. No arguments. Histories are available through get_nerfwatch_history. Scores and changes are rounded to two decimals; comparisons are within one suite version. A lower score is a lower score on this suite, not proof that a provider changed a model. Unofficial, not affiliated with OpenAI or Anthropic. Source: https://tokenusage.dev/nerfwatch

    Annotations the server declares

    • title: Nerf Watch scores
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false

This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.

Install

Pinned to version 1.0.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http nerfwatch https://tokenusage.dev/mcp/nerfwatch

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json
{
  "permissions": {
    "allow": [
      "mcp__nerfwatch__get_nerfwatch_scores",
      "mcp__nerfwatch__get_nerfwatch_history"
    ],
    "ask": [],
    "deny": []
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. It assumes the server is installed as nerfwatch. Review it before you use it.