Zero Review Plugin

This plugin uses software engineering discipline and test-driven verification to make coding agents produce high-quality code in one go, without human review or intervention.

It achieves this through structured engineering practice: every task goes through a mandatory design-thinking phase (architecture for greenfield, impact analysis for enhancements, hypothesis-driven diagnosis for bugs) before any code is written. This forces the agent to understand the problem deeply enough to get it right the first time. Then, test-plan-driven development (TPDD) defines what "correct" means before implementation — Must Have checkpoints, integration boundaries, and Forbidden Zone redlines give the agent concrete, verifiable success criteria it can check autonomously. Quality gates at every phase boundary replace the human reviewer: the agent self-assesses against objective criteria, and escalates only when something doesn't fit. The result is code that passes its own review — designed, implemented, reviewed, and verified — delivered without the human ever needing to read a line.

---

Installation

From Marketplace

/plugin marketplace add https://github.com/A7um/zero-review
/plugin install zero-review@atum-marketplace
/reload-plugins

Local Development

To test the plugin locally during development:

claude --plugin-dir /path/to/zero-review
/reload-plugins

After installation, the slash commands (/zero-review:dev, /zero-review:dev-new, /zero-review:dev-fix, /zero-review:dev-enhance, /zero-review:dev-add) become available, and the SubagentStart hook automatically injects skill context into subagents.

Capabilities

What	How
Autonomous delivery	Design → implement → self-review → verify → deliver, no human gates
Test-driven quality	TPDD defines success criteria before code is written; verification runs against real environments
Task-type workflows	4 paradigms (greenfield, enhancement, bugfix, addition) composed from 11 reusable phases
Parallel execution	Contract-based session isolation lets independent modules build in parallel
Anti-inertia design	Development philosophy + 8 named failure traps counter predictable agent mistakes
Cross-session learning	Per-project experience accumulation — conventions, pitfalls, patterns carry across sessions

Slash Commands

Command	Paradigm	When to Use
/zero-review:dev <task>	Auto-classify	Let the agent pick the right paradigm
/zero-review:dev-new <task>	dev/architecture-first	Greenfield — new project or system
/zero-review:dev-enhance <task>	enhancement/delta-design	Feature addition, behavior extension
/zero-review:dev-fix <task>	bugfix/hypothesis-driven	Defect, regression, incorrect behavior
/zero-review:dev-add <task>	addition/lightweight	Single function/component, fits existing architecture

Plugin Structure

plugins/zero-review/
├── .claude-plugin/plugin.json     # Plugin metadata
├── commands/                      # Slash command definitions
│   ├── dev.md                     #   /zero-review:dev — auto-classify
│   ├── dev-new.md                 #   /zero-review:dev-new — greenfield
│   ├── dev-enhance.md             #   /zero-review:dev-enhance — enhancement
│   ├── dev-fix.md                 #   /zero-review:dev-fix — bugfix
│   └── dev-add.md                 #   /zero-review:dev-add — addition
├── hooks/hooks.json               # SubagentStart hook — injects skill context
├── scripts/inject-dev-skill.sh    # Hook script — skill awareness for subagents
└── skills/auto-dev/              # The skill (auto-dev)
    ├── SKILL.md                   # Entry point — philosophy, classification, paradigms, traps
    ├── phases/                    # 11 reusable workflow steps + template
    ├── paradigms/                 # 4 task-type workflows + parallel-execution protocol
    │   ├── dev/                   #   architecture-first
    │   ├── enhancement/           #   delta-design
    │   ├── bugfix/                #   hypothesis-driven
    │   └── addition/              #   lightweight
    ├── principles/                # 8 design principles (module-depth, cohesion, etc.)
    ├── references/                # Lazy-loaded resources
    │   └── project-patterns/      #   Per-project experience (dynamic, gitignored)
    ├── scripts/match-project.sh   # Project experience matcher
    ├── config/defaults.json       # Configuration
    ├── examples/                  # Output format references
    └── CONTRIBUTING.md            # How to extend: add paradigms, compose workflows

How It Works

For the user: Type /zero-review:dev fix the login timeout bug or just describe the task. The plugin classifies it, loads the right paradigm, and the agent follows the phase sequence autonomously.

Design Principles

The skill embeds 8 software design principles (drawn from A Philosophy of Software Design) that are loaded selectively during architecture and code-review phases:

module-depth · information-hiding · abstraction-layers · cohesion-separation · error-handling · naming-obviousness · documentation · strategic-design

Contributing

See CONTRIBUTING.md for how to add new paradigms, compose existing ones, and configure output locations.