This skill should be used when implementing non-trivial software features that require planning, review, testing, and documentation. It orchestrates a team of specialized development agents (Groucho, Chico, Zeppo, Harpo) through a rigorous 7-phase workflow to ensure features are properly architected, implemented, reviewed, verified, and documented. Use this skill for multi-file features, new functionality, or any work where quality and thoroughness matter. Do not use for trivial single-line fixes or quick experimental prototypes.
Orchestrates a rigorous 7-phase workflow with specialized agents for implementing non-trivial features. Triggers when you need to implement multi-file features or quality-critical work requiring planning, review, testing, and documentation.
/plugin marketplace add schuyler/duck-soup/plugin install dev@duck-soupThis skill inherits all available tools. When active, it can use any tool Claude has access to.
assets/IMPLEMENTATION.mdassets/PLAN.mdassets/REFLECTION.mdassets/REQUIREMENTS.mdassets/TESTING.mdreferences/phase-1-gather-requirements.mdreferences/phase-2-plan.mdreferences/phase-3-implement.mdreferences/phase-4-review.mdreferences/phase-5-verify.mdreferences/phase-6-document.mdreferences/phase-7-reflect.mdreferences/session-resumption.mdscripts/init_session.sh1. Ambiguity means proceed strictly
When uncertain whether an approval todo applies, that uncertainty is the reason to apply it. The process exists precisely for situations where you're tempted to skip it.
2. Phase boundaries are strict
3. Step-by-step prevents overwhelm
One logical unit per step (max 50 lines). Wait for confirmation between steps.
Use TodoWrite to track all workflow progress:
Initialize phase todos at skill start
Expand phases just-in-time
Exactly ONE todo in_progress at a time
Approval todos enforce gates
Persist state to TODO.md
Agent coordination via shared todo list
All agents work on ONE shared todo list for complete visibility:
Agents create todos for their domain with [AgentName] prefix
Agents mark their own todos in_progress/completed
Main agent creates and completes implementation todos
Exactly ONE in_progress applies globally
docs/sessions/YYYYMMDD-HHMM-<slug>/
├── TODO.md (Persistent todo state, updated throughout)
├── REQUIREMENTS.md (Phase 1)
├── PLAN.md (Phase 2)
├── IMPLEMENTATION.md (Phase 3)
├── TESTING.md (Phase 5)
└── REFLECTION.md (Phase 7)
Initialize: scripts/init_session.sh <feature-slug>
TODO.md persistence:
Understand what the user needs.
Load references/phase-1-gather-requirements.md for details.
Create implementation plan aligned with existing patterns.
Load references/phase-2-plan.md for details.
Execute plan step-by-step.
Load references/phase-3-implement.md for details.
Verify implementation meets requirements.
Load references/phase-4-review.md for details.
Ensure implementation works correctly.
Load references/phase-5-verify.md for details.
Update documentation.
Load references/phase-6-document.md for details.
Record learnings.
Load references/phase-7-reflect.md for details.
If todo state becomes out of sync:
scripts/
init_session.sh - Create session directoryreferences/
phase-1-gather-requirements.mdphase-2-plan.mdphase-3-implement.mdphase-4-review.mdphase-5-verify.mdphase-6-document.mdphase-7-reflect.mdsession-resumption.mdassets/
Load phase references only when entering that phase.
If user provides session slug to resume:
docs/sessions/<session-slug>/TODO.md to reconstruct workflow stateIf user requests new feature implementation:
This skill should be used when the user asks to "create an agent", "add an agent", "write a subagent", "agent frontmatter", "when to use description", "agent examples", "agent tools", "agent colors", "autonomous agent", or needs guidance on agent structure, system prompts, triggering conditions, or agent development best practices for Claude Code plugins.
This skill should be used when the user asks to "create a slash command", "add a command", "write a custom command", "define command arguments", "use command frontmatter", "organize commands", "create command with file references", "interactive command", "use AskUserQuestion in command", or needs guidance on slash command structure, YAML frontmatter fields, dynamic arguments, bash execution in commands, user interaction patterns, or command development best practices for Claude Code.
This skill should be used when the user asks to "create a hook", "add a PreToolUse/PostToolUse/Stop hook", "validate tool use", "implement prompt-based hooks", "use ${CLAUDE_PLUGIN_ROOT}", "set up event-driven automation", "block dangerous commands", or mentions hook events (PreToolUse, PostToolUse, Stop, SubagentStop, SessionStart, SessionEnd, UserPromptSubmit, PreCompact, Notification). Provides comprehensive guidance for creating and implementing Claude Code plugin hooks with focus on advanced prompt-based hooks API.