@devcxl/opencode-cabbage

@devcxl/opencode-cabbage

A full-lifecycle development plugin for OpenCode, covering requirements, design, tasks, coding, testing, review, and release with automated orchestration and parallel subagents.

English | 简体中文

Installation

// opencode.json
{
  "plugin": ["@devcxl/opencode-cabbage"]
}

Once started, the plugin automatically injects 7 slash commands, 9 flow skills, 6 agents, and 1 goal tool.

Command Overview

Command	Stage	Output
/setup	Setup	docs/ directory structure, Project Profile, CI/release workflow validation
/requirements	Requirements	PRD → docs/prd/ + Draft Parent Issue
/design	Design	Technical specification + necessary ADR → docs/dev/specs/ + docs/adr/
/tasks	Task decomposition	DAG tasks + Sub Issues (GitHub 权威源)
/code	Coding	Task + PR (TDD enforced by CI)
/review	Review	Dual-axis review + merge
/release	⚠️ Manual release	Version proposal → Release PR → tag push → workflow monitor

Quick Start

# 1. Install
npm install @devcxl/opencode-cabbage

# 2. Add the plugin to opencode.json
# { "plugin": ["@devcxl/opencode-cabbage"] }

# 3. Run in OpenCode
# /setup → /requirements → @dev-lifecycle

Two Modes

• Manual mode — Run each command sequentially for fine-grained control
• Automatic mode — Once the requirements are confirmed, enter @dev-lifecycle to automatically complete the remaining workflow

Architecture

src/                          # Thin TypeScript layer
├── index.ts                  # Plugin entry point
├── plugin.ts                 # Package root resolution
├── plugin/                   # Loaders + goal tool
│   ├── server.ts             # Main factory: injects skills/commands/agents, goal tool, context, auto-continuation
│   ├── goal.ts               # goal tool (status + goal-verify authorization)
│   ├── commands.ts           # Command loader
│   ├── skills.ts             # Skill loader
│   ├── prompts.ts            # Prompt loader
│   ├── bootstrap.ts          # Startup guidance
│   ├── agents.ts             # Agent injection
│   ├── shell.ts              # Agent shell env isolation
│   └── prompt-lint.ts        # Prompt asset consistency checks
└── kernel/                   # Minimal supporting modules
    ├── context.ts            # root CONTEXT.md discovery/injection
    ├── profile.ts            # AGENTS.md Project Profile parsing
    ├── session-index.ts      # Flow→session index (goal binding + auto-resume)
    ├── permission.ts         # Permission matching semantics
    └── mutex.ts              # Keyed async mutex

assets/                       # Runtime assets (pure Prompt flows)
├── commands/                 # 7 slash commands
├── skills/                   # 9 flow-* skills (Prompt-driven, direct git/gh)
├── agents/                   # 6 agent definitions
└── prompts/                  # Guidance prompts and templates

Documentation

Document	Link
Quick Start	docs/guides/quickstart.md
Configuration Guide	docs/guides/configuration.md
Usage Guide	docs/guides/usage.md
Architecture Overview	docs/guides/architecture.md
Contributing Guide	docs/dev/guides/contributing.md

License

MIT