Crucible
Run a structured, multi-persona corporate-style review on your project before you ship, with explicit aim alignment and a synthesized verdict.
Overview
Crucible is a plugin marketplace for the Ship phase that runs your project through a five-stage persona-based review pipeline and returns a synthesized score and verdict aligned to your stated aims.
What is this marketplace?
- Five-stage pipeline: Profiler → peer review → departments → leadership aim alignment → Aggregator verdict
- Profiler interviews you for aims and casts a 4–8 member committee from a 23-persona library
- Peers review at code level; departments hunt cross-cutting gaps; leadership grades aim alignment
- Aggregator synthesizes a final score and holistic verdict (Opus reasoning in metadata)
- Single-plugin Claude Code marketplace bundle (v0.1.2, MIT) from the crucible-cc repo
- Marketplace bundles 1 plugin (crucible v0.1.2)
- 5-stage review pipeline with a 23-persona library
- Profiler casts a 4–8 persona review committee per run
Community signal: 4 GitHub stars.
What problem does it solve?
Solo builders often ship without structured review, so code gaps and drift from original intent show up only after users or production break.
Who is it for?
Indie and solo developers on Claude Code who want a repeatable, aim-aware review ritual on a real codebase before release or a major merge.
Skip if: Teams that only need fast inline linting, or greenfield repos with nothing meaningful for a Profiler to read and no stated aims to grade against.
What do I get? / Deliverables
After you run Crucible, you get a committee-style review across code, gaps, and aim alignment, plus an Aggregator verdict you can use to fix issues or confidently ship.
- Staged peer, department, and leadership review outputs tied to your aims
- Final synthesized score and holistic verdict from the Aggregator
- Actionable gap and alignment signal to guide fixes before ship
Plugins in this marketplace
1 plugin — install individually after you add the marketplace.
Recommended Marketplaces
Journey fit
Crucible is a post-build quality gate: it profiles the repo, runs staged reviews, and outputs a scored verdict—work that belongs in Ship when you are validating readiness to release. The plugin is explicitly a code-review pipeline with peer-level review, gap hunting, and leadership grading, which maps directly to the review subphase rather than security scanning or launch perf checks.
How it compares
Multi-stage persona review pipeline plugin for Claude Code, not a single MCP tool or one-shot static analyzer.
Common Questions / FAQ
Who is Crucible for?
Crucible is for solo and indie builders using Claude Code who want structured, multi-persona code and aim-alignment review before they ship or merge significant work.
When should I use Crucible?
Use Crucible in the Ship phase when you have a project or branch ready for review and you want gaps, code-level feedback, and leadership-style aim grading rolled into one verdict.
How do I add Crucible to my agent?
Add the Crucible marketplace from hazarsozer/Crucible-cc in Claude Code, install the Crucible plugin from the marketplace manifest, then invoke the plugin workflow against the project you want reviewed.