Skill UI

This advanced skill analyzes an existing codebase to identify architectural patterns, dependencies, and technical stacks. It recommends tailored Claude Code automations—including Hooks, Subagents, Skills, Plugins, and MCP Servers—to optimize development workflows, improve code consistency, and enhance overall project scalability. Use this when setting up a new project or looking to refine existing Claude Code configurations.

Guides creation of Cursor-specific subagents with isolated contexts, structured prompts, and frontmatter so multi-step workflows, verifiers, or specialists follow Cursor’s directories and delegation rules.

Guides orchestrating multiple Task tool subagents with run_in_background for concurrent work, covering when to trigger parallelism, how to craft prompts, launch in one message, synchronize results, and integrate TodoWrite status updates.

Runs Semgrep static analysis with auto language detection, parallel Task subagents, Semgrep Pro checks, and configurable “run all” or “important only” modes to audit codebases for vulnerabilities and bug patterns.

Provides a structured workflow for defining agent-agnostic subagents with isolated context, covering purpose definition, metadata configuration, prompt writing, and common specialist patterns for verification, debugging, security auditing, and review.

Guides execution of implementation plans by dispatching isolated subagents per task, enforcing a spec-first then quality review loop, and preserving overall coordination context for faster, high-quality iterations.

Performs a full, multi-dimensional SEO audit by crawling up to 500 pages and delegating tasks to specialized subagents. It assesses technical health, content quality (E-E-A-T), performance (CWV), local signals, and more. The tool culminates in a comprehensive SEO Health Score and a prioritized, actionable optimization plan, often delivered as a professional PDF report.

Performs a full GEO audit of any website by discovering business type, crawling prioritized pages, and delegating parallel subagents to evaluate AI visibility, platform readiness, technical infrastructure, EEAT content, and schema markup, then outputs a composite GEO score with prioritized actions.

The /hub:spawn command launches N subagents that work on the same task simultaneously. Each agent operates within an isolated git worktree, ensuring true parallelism and resource separation. Users can optionally specify templates (e.g., `optimizer`, `refactorer`) to guide agents toward specific strategies for improving code quality, performance, or test coverage. Ideal for complex software engineering tasks requiring diverse, competing approaches.

This tool generates production-grade, spec-compliant agent definition files (.md) adhering to the Anthropic 2026 schema. It supports both the creation of new custom subagents and the validation of existing agent files against marketplace compliance rules. Use it when building complex agent architectures, reviewing agent quality, or ensuring agents meet high industry standards.

A software development methodology that delegates independent tasks to isolated subagents in the current session, ensuring high quality through two-stage reviews focusing on spec compliance and code quality.

Cavecrew provides a set of specialized subagents (investigator, builder, reviewer) designed for efficient code development delegation. Unlike vanilla agents, Cavecrew compresses output results, drastically reducing the token usage injected back into the main context. This allows users to maintain context over much longer, complex coding sessions, effectively preventing context exhaustion during large-scale refactoring, debugging, or multi-step feature development.