ZeroWidth Caliper
ai.zerowidthv1.0.0Updated Oct 7, 2026
Build datasets and rubrics, then run evals that score your AI in Caliper.
Overview
AI-generated overview
Lets an assistant build datasets and rubrics and run evals that score AI outputs in ZeroWidth Caliper.
- What it does
- This remote MCP server connects an assistant to ZeroWidth Caliper, where it can build datasets and rubrics and then run evaluations that score AI behavior. The registry description states it lets you "Build datasets and rubrics, then run evals that score your AI in Caliper." No individual tools are listed in the manifest, so the exact operations are not documented here.
- When to use it
- Consider it when you want an assistant to help assemble evaluation datasets and scoring rubrics and to trigger or review eval runs against your AI in Caliper, rather than doing that work by hand in the Caliper interface.
- Requirements
- A remote streamable HTTP endpoint at It requires an Authorization header carrying a Bearer token: a workspace API key or personal access token from accounts.zerowidth.ai. No packages or environment variables are declared.
Before you install
The Authorization header carries a workspace API key or personal access token, so treat it as a secret and avoid exposing it in logs or shared configs. The server builds datasets and rubrics and runs evals, which may create or change data in your Caliper workspace; confirm what it writes before relying on results.
Installation
In SourceWeft
- Open ZeroWidth Caliper in the dashboard and add it to a workspace.
- Enable the server for the chats that should use its tools.
Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.
Other MCP clients
Add this to your client's mcpServers config.
{
"mcpServers": {
"caliper": {
"type": "http",
"url": "https://api.zerowidth.ai/mcp/x/caliper"
}
}
}Tools
0Tool metadata has not been indexed yet.
Version history
1- v1.0.0LatestOct 7, 2026