@orangekame3/pi-qcaleval

pi extension that sends calibration plots to a calibration-evaluation VLM such as NVIDIA Ising-Calibration and returns a structured diagnosis

Packages

Package details

extensionskill

Install @orangekame3/pi-qcaleval from npm and Pi will load the resources declared by the package manifest.

npm repo home report

$ pi install npm:@orangekame3/pi-qcaleval

Package: @orangekame3/pi-qcaleval
Version: 0.0.1
Published: Oct 9, 2026
Downloads: 1,188/mo · 71/wk
Author: orangekame3
License: Apache-2.0
Types: extension, skill
Size: 26 KB
Dependencies: 0 dependencies · 3 peers

Pi manifest JSON

{
  "skills": [
    "./skills"
  ],
  "extensions": [
    "./extensions/qcaleval.ts"
  ]
}

Security note

Pi packages can execute code and influence agent behavior. Review the source before installing third-party packages.

README

pi-qcaleval

A pi extension that sends superconducting qubit calibration plots to a dedicated calibration-evaluation VLM and returns a structured diagnosis. The default model is NVIDIA Ising-Calibration served by vLLM.

The planner model keeps the conversation; qcal_evaluate is the tool it calls whenever a calibration figure needs a fit-quality judgement, parameter extraction, or a success verdict. The extension has no QDash dependency, so it works from the pi CLI and from the QDash Copilot runtime alike.

Tool

qcal_evaluate(context, question?, images?, max_images?) returns a JSON object with technical_description, experimental_conclusion, experimental_significance, fit_quality, parameter_extraction, and success. The parsed object is also stored in the tool result's details.diagnosis; when the model did not answer with JSON, details.raw holds its text instead.

images are base64 image blocks. A host may fill them in from the most recent figures of the conversation, so the planner does not have to pass image bytes; max_images (default 1, up to 4) says how many of the newest figures to supply, for example 2 when an expected reference figure precedes the measured result.

Configuration

Variable	Purpose
`QCALEVAL_MODEL`	`provider/modelId` of the evaluation model in pi's `models.json` (default: `vllm/nvidia/Ising-Calibration-1.5-31B-NVFP4`)
`QCALEVAL_MAX_TOKENS`	Output budget for one evaluation (default: `2048`)

The provider must be declared in pi's models.json with its base URL and credential, for example:

{
  "providers": {
    "vllm": {
      "baseUrl": "https://llm.example/v1",
      "api": "openai-completions",
      "apiKey": "$VLLM_API_KEY",
      "models": [{ "id": "nvidia/Ising-Calibration-1.5-31B-NVFP4", "contextWindow": 16384, "maxTokens": 2048, "input": ["text", "image"] }]
    }
  }
}

Only the plot, a short system prompt, and the context text are sent, so the evaluation fits comfortably in a 16k-token deployment.

Installation

pi install npm:@orangekame3/pi-qcaleval

Releases are tagged vX.Y.Z and published to npm as latest; every push to main also publishes a beta prerelease (npm:@orangekame3/pi-qcaleval@beta) for hosts that want an unreleased build. Installing from git (pi install git:github.com/orangekame3/pi-qcaleval) also works.

For development against the QDash Agent Runtime, clone this repository under agent-runtime/extensions/ in the QDash checkout and start the runtime with compose.extensions.yaml; see the QDash Copilot architecture docs.

Development

bun install
bun run check   # tsc --noEmit
bun run test    # node --test

License

Apache-2.0