LLM & Agents

Use when creating new skills, editing existing skills, or verifying skills work before deployment - applies TDD to process documentation by testing with subagents before writing, iterating until bulletproof against rationalization

Use when completing tasks, implementing major features, or before merging to verify work meets requirements - dispatches superpowers:code-reviewer subagent to review implementation against plan or requirements before proceeding

Expert guidance for GRPO/RL fine-tuning with TRL for reasoning and task-specific model training

Framework for building LLM-powered applications with agents, chains, and RAG. Supports multiple providers (OpenAI, Anthropic, Google), 500+ integrations, ReAct agents, tool calling, memory management, and vector store retrieval. Use for building chatbots, question-answering systems, autonomous agents, or RAG applications. Best for rapid prototyping and production deployments.

Simple Preference Optimization for LLM alignment. Reference-free alternative to DPO with better performance (+6.4 points on AlpacaEval 2.0). No reference model needed, more efficient than DPO. Use for preference alignment when want simpler, faster training than DPO/PPO.

Multi-agent coordination patterns for OpenCode swarm workflows. Use when working on complex tasks that benefit from parallelization, when coordinating multiple agents, or when managing task decomposition. Do NOT use for simple single-agent tasks.

Multi-agent coordination patterns for OpenCode swarm workflows. Use when working on complex tasks that benefit from parallelization, when coordinating multiple agents, or when managing task decomposition. Do NOT use for simple single-agent tasks.

This skill should be used when the user asks to "create a pull request", "create PR", "open PR", "create an issue", "file an issue", "create a GitHub issue", "create a Claude Code issue", "report a bug in Claude Code", "create a Codex issue", "report a bug in Codex CLI", "create a discussion", "start a GitHub discussion", or mentions OSS contribution workflows.

多AI协作编排技能 - 自动协调Claude、Codex、Gemini完成需求分析→代码生成→代码审查工作流。 【核心触发场景】 - 开发任务、写代码、实现功能、编程任务 - 新功能开发、功能实现、代码实现 - 重构、优化代码、代码改进 - Bug修复、问题修复、错误处理 - 系统设计、架构设计、模块设计 【显式触发词】 - "AI编排"、"ai编排"、"多AI协作"、"三引擎"、"三AI" - "codex"、"gemini"、"kim team"、"kim-team" - "协作开发"、"编排流程" 【任务类型触发】 - 实现XXX功能、开发XXX模块、写一个XXX - 创建XXX、构建XXX、搭建XXX - 添加XXX功能、新增XXX特性 - 修改XXX、更新XXX、升级XXX - 修复XXX、解决XXX问题 - 重构XXX、优化XXX性能 - 设计XXX架构、规划XXX系统 【技术任务触发】 - API开发、接口实现、后端开发、前端开发 - 数据库设计、表结构、CRUD操作 - 用户认证、登录注册、权限控制 - 文件处理、数据导入导出 - 第三方集成、SDK接入

This skill should be used when the user asks to "update README", "update context files", "init context", "create CLAUDE.md", "update CLAUDE.md", "update AGENTS.md", "update DOCS.md", "generate documentation", "API documentation", or mentions project documentation, context files, or markdown documentation workflows.

Use this skill when the user asks to add documentation, add docs, add references, or install documentation about Neon. Adds Neon best practices reference links to project AI documentation (CLAUDE.md, AGENTS.md, or Cursor rules). Does not install packages or modify code.

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Use this skill when creating, improving, or optimizing prompts for Claude. Applies Anthropic's best practices for prompt engineering including clarity, structure, consistency, hallucination reduction, and security. Useful when users request help with writing prompts, improving existing prompts, reducing errors, increasing consistency, or implementing specific techniques like chain-of-thought, multishot prompting, or XML structuring.

Custom agent and workflow creation specialist

Creates custom agents, workflows, and templates for BMAD. Extends BMAD functionality with domain-specific components. Trigger keywords - create agent, create workflow, custom skill, extend BMAD, new template, customize, scaffold skill

Create or augment skills and documentation. Use when writing new skills, updating existing skills, or modifying project documentation like AGENTS.md.

Persistent graph-based memory for AI agents via KIP (Knowledge Interaction Protocol). Provides structured knowledge storage (concepts, propositions), retrieval (KQL queries), schema discovery (META), and memory metabolism. Use when: (1) remembering user preferences, identities, or relationships across sessions, (2) storing conversation summaries or episodic events, (3) building and querying knowledge graphs, (4) the user says "remember this", "what do you know about me", or asks about past conversations, (5) needing to maintain context continuity across sessions. Requires HTTP access to a KIP backend (anda_cognitive_nexus_server).

Performs code quality checks and reviews for ComposeWebView. Validates expect/actual implementations, KDoc coverage, Spotless formatting, and multiplatform patterns. Use when reviewing PRs, checking code quality, or validating new features.

This skill should be used when the user asks to "write tests", "test strategy", "coverage", "unit test", "integration test", or needs testing guidance. Provides testing methodology and patterns.

Configure TDD enforcement via pre-commit hooks and CI coverage gates. Use when setting up test-first development workflow, adding coverage gates, or enforcing TDD practices.