A Claude Code skill ecosystem for structured software development workflows.
| Skill | Purpose | Triggers |
|---|---|---|
| brainstorming | Vague idea → research direction | "I want to build...", "How might we...", unclear requirements |
| research | Evaluate options, gather context | Package selection, architecture decisions, gray areas |
| planning | Break work into phases and tasks | Feature specs, bug fixes requiring multiple changes |
| executing | Run tasks with subagents | Phase plan ready, independent tasks to parallelize |
| reviewing | Verify work against spec | Implementation complete, before merge |
| debugging | Root cause analysis | Errors, test failures, unexpected behavior |
| prompt-leverage | Structure effective prompts | Delegating to subagents, complex task handoffs |
brainstorming → research → planning → executing → reviewing
↑ ↓
debugging ←────────────────┘
git clone https://github.com/duchm1606/swekit.git ~/.claude/skills/duckSkills are available in all your projects as duck:*.
cd your-project
git submodule add https://github.com/duchm1606/swekit.git .claude/skills/duckcd ~/.claude/skills/duck # or your install location
git pullSkills are invoked automatically when Claude detects matching triggers, or manually:
/duck:brainstorming # Start with vague idea
/duck:research # Evaluate packages/approaches
/duck:planning # Create phase plan
/duck:executing # Run with subagents
/duck:reviewing # Verify against spec
/duck:debugging # Root cause analysis
Built by studying and adapting from:
- GET SHIT DONE — context engineering, deep brainstorming
- Superpowers — TDD discipline, subagent isolation, systematic debugging
- Khuym — beads task graph, validation gates
- ClaudeKit Engineer — multi-agent research, domain skills library