<div align="center">

<img src="assets/hero.png" alt="Crucible" width="520" />

# Crucible

**Codebase-level adversarial review by a panel of frontier models.**

A Claude Code skill that walks your code piece-by-piece and puts every file under simultaneous pressure from a panel of structurally different models, then aggregates the findings into a single severity-ranked report that Claude itself verifies before you see it.

[Install](#install) · [How it works](#how-it-works) · [Cost](#cost) · [Modes](#modes) · [Sample report](#sample-report)

</div>

---

## What it is

`/crucible` is a [Claude Code](https://claude.com/claude-code) slash-command skill. You drop the folder into `~/.claude/skills/`, set one env var, and from inside any project you can run:

``` /crucible # review the current branch's diff /crucible --all # review the whole repo /crucible --paths "src/api/**/*.ts" # review a glob /crucible --diff main...HEAD # review a specific range /crucible --verify # also EXECUTE a repro for runtime-flagged findings ```

Behind the scenes, Claude:

1. Resolves the file list and prints a pre-flight (files, models, est. cost). 2. Loads a panel of four current SOTA paid models from OpenRouter, each from a different vendor family (e.g. DeepSeek, Google, Moonshot, MiniMax). 3. Reviews every in-scope file through the panel: pass 1 finds, pass 2 validates, pass 3 consolidates with severity ranks. 4. Runs one cross-file architectural meta-pass to catch repeated anti-patterns, missing layers, and coupling smells. 5. **(opt-in `--verify`) Executes a repro for the runtime bugs.** Some bugs only exist when the code runs — a retry the surrounding rate-limit always rejects, a `2>/dev/null` that hides a device error. For each finding the panel tags *runtime-checkable*, a model writes a minimal repro harness, Crucible runs it in a **locked sandbox**, and findings that actually reproduce are promote