AI Engineering

Plan Mode creates detailed implementation strategies before coding, including requirements, design, and task breakdown, and waits for explicit approval before starting work.

Role Debate triggers a structured multi-role discussion in Claude Code to explore trade-offs and reach integrated conclusions. Use it when you need to balance competing perspectives like security vs performance or architecture choices.

Role Help guides users in choosing the right Claude Code role for their situation, covering security, architecture, performance, frontend, mobile, review, QA, and multi-role comparisons.

Role Switching lets a Claude Code agent adopt expert personas like security auditor, performance optimizer, code reviewer, architect, or QA engineer to perform specialized, evidence-based analysis. Use it when you need a focused audit, review, or optimization with structured professional output.

Add Upstream Skill adds a remote skill entry to the active workspace without copying its source. Use it to keep your library shaped like your own while preserving upstream metadata and choosing shelf, branch, and reason.

Ask Questions If Underspecified prompts an AI agent to pause and ask 1-5 targeted clarifying questions when a request is ambiguous, then confirms assumptions before proceeding so it doesn't start wrong work.

Best Practices transforms vague prompts into optimized Claude Code prompts by adding verification, specific context, constraints, and proper phasing. Invoke with /best-practices when you need to improve a prompt.

Browse And Evaluate uses the ai-agent-skills CLI to search and inspect skills before installing them. Use it to keep context usage low with --fields and --dry-run.

Curate a Team Library builds a managed team skills library for a real stack by mapping work to shelves, browsing before curating, and writing meaningful whyHere notes. Use it when setting up a shared library that teammates or agents can browse, trust, and install.

LLM Application Development provides patterns for prompt engineering, RAG pipelines, and LLM API integration. Use it to build AI-powered features, chatbots, and automation.

Migrate Skills Between Libraries moves skills between AI agent library workspaces using the ai-agent-skills CLI, handling export, import, collision checks, and dependency migration. Use it when upgrading from a personal to a team library or consolidating workspaces.

Update Installed Skills keeps installed agent skills current without breaking the workflow or silently overwriting local customizations. Use it when syncing or updating previously installed skills to their latest version, always with a dry-run before applying.