AI Engineering

Arize Annotation creates and manages annotation configs (label schemas) and annotation queues (human review workflows) on Arize, and applies human annotations to project spans via the Python SDK. Use it when setting up human feedback, label schemas, or bulk span annotations.

Arize Evaluator creates and runs LLM-as-judge evaluators on Arize, including setting up evaluator templates, running evaluations on spans or experiments, and managing tasks and column mappings. Use it when working with hallucination, faithfulness, correctness, relevance, or other LLM quality metrics.

Generate AI Agent Instructions creates tailored instruction files for AI coding agents using AgentRC's instructions command, including .github/copilot-instructions.md and scoped per-area .instructions.md files. Use it after an assessment to close gaps in AI tooling.

Build MCP App adds interactive UI widgets to an MCP server so tools can render forms, pickers, confirmations, dashboards, and progress indicators inline in chat. Use it after build-mcp-server when you need structured input, visual output, or explicit user confirmation.

Claude Automation Recommender analyzes a codebase to recommend tailored Claude Code automations such as hooks, subagents, skills, plugins, and MCP servers. Use it when setting up Claude Code for a project or optimizing existing workflows.

Claude MD Improver audits and improves CLAUDE.md files across a codebase so Claude Code has accurate project context. Use it when checking, updating, or maintaining CLAUDE.md files or project memory.

Example Command is a sample user-invoked Claude Code skill that demonstrates the skills/<name>/SKILL.md layout and frontmatter options.

Math Olympiad Solver is a multi-agent workflow for solving olympiad-style math problems and verifying proofs through adversarial passes. Use it for IMO, Putnam, USAMO, or AIME problems when rigorous solutions and gap detection matter.

Skill Template provides a reference template for creating Claude Code plugin skills. It activates when users ask to demonstrate skills, show skill format, or discuss skill development patterns.

Writing Hookify Rules guides creating hookify rule files with YAML frontmatter to define patterns and messages for Claude Code events. Use it when asked to create, write, or configure hookify rules, or need syntax and pattern guidance.

Agent Development guides the creation of autonomous agents for Claude Code plugins, covering agent structure, triggering conditions, system prompts, and best practices. Use it when building subagents or configuring agent frontmatter.

Build MCP Server guides developers through designing and building MCP servers that work with Claude, from discovery questions to deployment recommendations. Use it when asked to build an MCP server, create an MCP integration, wrap an API for Claude, or expose tools to Claude.