pitcany

Review Calibration — Development skill for Claude Code

Development community

Review probability quality — separates discrimination from calibration, checks reliability curves, calibration-in-the-large and slope, subgroup calibration, and calibration method choice.

How to install Review Calibration

Installs to ~/.claude/skills/pitcany-statistical-agent-skills-review-calibration/SKILL.md

Terminal

mkdir -p ~/.claude/skills/pitcany-statistical-agent-skills-review-calibration && curl -fsSL https://raw.githubusercontent.com/pitcany/statistical-agent-skills/HEAD/commands/review-calibration.md -o ~/.claude/skills/pitcany-statistical-agent-skills-review-calibration/SKILL.md

Restart Claude Code, or start a new session, for it to be picked up.

What Review Calibration does

description: Review probability quality — separates discrimination from calibration, checks reliability curves, calibration-in-the-large and slope, subgroup calibration, and calibration method choice.

Calibration and ranking review

Target: $ARGUMENTS

What to do

Invoke the calibration-and-ranking skill.
State which claim is actually being made — "we can rank" or "the probabilities are right" — and which evidence was supplied for it. These are orthogonal axes.
AUROC

Alternatives in Development

Cohort Analysis — Cohort retention curves, feature adoption, and segment insights 7.8k ★
Flow Parallel — Decompose and execute large changes, migrations, or multi-issue fixes in parallel with quality gates 2.8k ★
Claude Gemini Bridge — Intelligent integration between Claude Code and Google Gemini for large-scale code analysis 386 ★