Stats
Actions
Tags
From clippy
This skill should be used when the user says "clippy", "use clippy", "with clippy", or asks to "build a feature", "investigate the codebase", "refactor this", "implement X", or requests evidence-based development requiring investigation before implementation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/clippy:composerThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
**Instructions for AI assistant.** User triggers when requesting codebase investigation or feature development. When triggered, first response MUST follow PROTOCOL ENTRY; do not skip to implementation or unstructured investigation.
Instructions for AI assistant. User triggers when requesting codebase investigation or feature development. When triggered, first response MUST follow PROTOCOL ENTRY; do not skip to implementation or unstructured investigation.
TRIGGER: User requested investigation, feature, or refactor; OR user explicitly invoked this protocol.
BEFORE first response to a CLIPPY-scope request:
clippy:investigate-design now.Bypassing protocol: Starting implementation or investigating without tracker + menu in same response.
CORE PRINCIPLE: AI judgment augmented by systematic verification discipline. Checklists ensure minimum coverage; AI's own observations carry equal weight.
APPROACH: Three-phase workflow:
clippy:investigate-designclippy:implementclippy:verifyLOOP STRUCTURE: INVESTIGATE & DESIGN continues until user chooses "i" AND [READY] is confirmed.
PROTOCOL ACTIVE UNTIL EXPLICIT EXIT: Protocol remains active until user explicitly exits via "d" (post-implementation or VERIFY). No implicit exit. Menu signals protocol is still active; without menu, user cannot choose next action.
DEFAULT IN INVESTIGATE & DESIGN:
Detect current phase from context:
clippy:investigate-designCompletion signals from phase skills:
Phase skills end with specific signals:
🔚 IMPLEMENT PHASE COMPLETE — implementation done🔚 IMPLEMENTATION ABORTED — readiness failed or major scope discovered🔚 VERIFY PHASE COMPLETE — verification round doneNote: Investigation skill does NOT emit per-cycle completion signals. It includes the menu inline in each cycle response. The terminal investigation signal is only used internally when transitioning to implement.
When completion signal detected → composer takes over, shows appropriate menu.
Every response during protocol:
📌 This turn: (2-4 lines: what changed, next step, blocker)Phase-specific output:
clippy:investigate-design)Format: Show ONLY options for current phase. --- before/after. Title: ☰ **Menu:**. Bold command letter. Free-form invitation at end.
---
☰ **Menu:**
- 🔍 **c** - continue: More investigation/design iteration
- 🚀 **i** - implement: Start implementation (requires [READY])
- 🤖 **auto** - switch to autonomous: Hand off to autopilot (requires OMC + [READY])
🙋 N user decision(s) pending — see tracker ← (only when USER DECISIONS non-empty)
... or anything else?
---
---
☰ **Menu:**
- ✅ **v** - verify: Run verification
- ✨ **d** - done: Exit protocol (skip verification)
... or anything else?
---
---
☰ **Menu:**
- ➡️ **c** - continue: Another verification round
- ✨ **d** - done: Feature complete
... or anything else?
---
c - continue:
clippy:investigate-design for another cycle. More investigation/design iteration (widens scope, deepens investigation — NOT readiness to implement).clippy:verify for another round.When routing "c" during INVESTIGATE & DESIGN:
i - implement:
BEFORE activating clippy:implement:
clippy:implement.v - verify:
clippy:verify. Run verification once; then show VERIFY menu (c / d).auto - switch to autonomous:
clippy:autopilot with ALL current state intact.
The investigation tracker, design decisions, constraints — everything
done so far carries over. Autopilot continues from the current state,
not from scratch.BEFORE activating clippy:autopilot:
.ai/ (tracker, design artifacts,
constraints). Autopilot reads from disk.clippy:autopilot with the current task context. Autopilot
detects existing .ai/ state and continues from where interactive left off.d - done:
PRINCIPLE: Free-form feedback stays within protocol.
When responding to free-form feedback:
MENU ENFORCEMENT (BEFORE ending any response):
START → investigate-design (loop via "c")
↓ "i" (requires [READY])
implement (runs to completion)
↓ auto
Post-implementation menu (v / d)
↓ "v"
verify (loop via "c")
↓ "d"
EXIT
Implementation runs to completion. No intermediate menus during implementation. The implement skill executes all planned steps and returns a completion signal. User can interrupt with free-form input at any time (implement skill pauses and responds).
Auto-transition: When implementation completes, show post-implementation menu. Default suggestion: "v" (verify).
Implementation abort: If implement skill emits 🔚 IMPLEMENTATION ABORTED, re-enter INVESTIGATE & DESIGN with existing tracker. Show investigation menu.
Back to investigation: If user requests returning to investigation post-implementation (via free-form), re-enter INVESTIGATE & DESIGN with existing tracker.
Last Updated: April 2026 (v2.1 - Core principle: augmentation framing. Previous: v2.0.)
npx claudepluginhub gunther-schulz/coding-clippyMulti-AI implementation with structured development workflow. Enforces pre-development state checks and detects whether the task is dev or knowledge work to route appropriately.
Guides multi-phase feature development with research, planning, implementation, and review phases. Use for complex features touching >5 files or requiring architecture decisions.
Coding workflow covering discovery, planning, implementation, and verification. Invoke whenever task involves any interaction with code — writing, modifying, debugging, refactoring, or understanding codebases. Runs discovery protocol before language-specific skills engage.