메뉴
End-to-end feature development orchestrator. Composes /research, /brainstorm, /plan, and /execute into a single workflow. Use when building a new feature from scratch. Triggers on "build a feature", "implement X from scratch", "full feature development for Y".
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
Agentically test a specific feature by analyzing the branch diff, building a test plan, and driving the app through agent-browser. Builds on /login and /error skills. Caches successful playbooks to .ai/playbooks/ for reuse.
Research-driven brainstorming and design for a feature. Use when designing a new feature, exploring approaches, or making design decisions. Automatically researches competitors before asking design questions, then hands off to /plan for implementation. Triggers on "brainstorm", "design", "design a feature", "plan a feature", "how should we build X", or any request to explore approaches for a feature.
Author, edit, or extend the Carbon documentation site at `docs` (a built Fumadocs + Next.js app). Use whenever creating or changing reader-facing docs for Carbon: editorial Guide chapters, Reference/entity pages, the docs IA, or "document this feature" requests — anything destined for the docs site. Covers the grounded-in-source authoring workflow, the flow-based Guide architecture, the real MDX components for each surface, the warm-paper house style, and the build verification loop. Trigger even if the user doesn't say "docs" but clearly wants explanatory, reader-facing written material about how Carbon works.
Supervised loop conductor — drive a single work item (a bug, a usability tweak, a small feature) through a doer→gate→keep-or-revert→ledger cycle to a gated PR, while the human watches. Use when asked to "conduct", "build", "fix", or "loop on" a specific item with the conductor. Supervised only — not autonomous/overnight (not yet supported). Backed by @carbon/harness + the @carbon/checks checks.
Execute implementation plans task by task with verification. Use after /plan to implement the feature. Follows each task exactly, runs verifications, commits frequently. Triggers on "execute the plan", "implement the plan", "run the plan", or after /plan approval.
Survey the Carbon codebase as a senior advisor and produce prioritized, self-contained implementation plans for OTHER models/agents to execute. Strictly read-only on source code — never implements, fixes, or refactors anything itself. Use when asked to audit Carbon, find improvement opportunities (bugs, security, performance, test coverage, tech debt, migrations, DX), suggest features or where to take the project next (roadmap, product direction), or generate handoff plans for another agent to implement.
| name | feature |
| description | End-to-end feature development orchestrator. Composes /research, /brainstorm, /plan, and /execute into a single workflow. Use when building a new feature from scratch. Triggers on "build a feature", "implement X from scratch", "full feature development for Y". |
Orchestrates the full feature development lifecycle by composing smaller skills:
/research → /brainstorm → /plan → /execute → /verify
Announce at start: "I'm using the feature skill to orchestrate end-to-end development."
Use /feature when... | Use individual skills when... |
|---|---|
| Building something new from scratch | You only need one phase |
| Want the full research-to-implementation flow | Design already exists |
| Don't want to invoke each skill manually | Plan already exists |
/research)Identify the domain and research competitors:
.ai/scratch/research/[feature].md/brainstorm)Design with research in hand:
.ai/scratch/research/[feature]-design.md/plan)Create implementation plan:
.ai/scratch/plans/[feature]-plan.md/execute)Implement the plan:
/verify)Test in browser:
User: /feature lot tracking for inventory
Claude: I'm using the feature skill to orchestrate end-to-end development.
Phase 1: Research
[Invokes /research inventory lot tracking]
[Saves to .ai/scratch/research/lot-tracking.md]
Phase 2: Brainstorm
[Invokes /brainstorm with research context]
[Presents design spec]
[Waits for approval]
Phase 3: Plan
[Invokes /plan]
[Presents implementation plan]
[Waits for approval]
Phase 4: Execute
[Invokes /execute]
[Implements each task]
Phase 5: Verify
[Invokes /verify]
[Tests feature]
| Phase | Artifact | Location |
|---|---|---|
| Research | Competitor analysis | .ai/scratch/research/[feature].md |
| Brainstorm | Design spec | .ai/scratch/research/[feature]-design.md |
| Plan | Implementation plan | .ai/scratch/plans/[feature]-plan.md |
| Execute | Working code | Feature branch |
| Verify | Test evidence | Screenshots/logs |
Each phase uses a standalone skill you can invoke directly:
/research [topic] — Survey competitor best practices/brainstorm [feature] — Research-driven design/plan [feature] — Create implementation plan/execute [feature] — Implement from plan/verify — Test feature in browser/self-review — Review before PRIf you already have artifacts, skip to the appropriate phase:
/brainstorm/plan/execute