LLM & Agents

Use when developing PRPM (Prompt Package Manager) - comprehensive knowledge base covering architecture, format conversion, package types, collections, quality standards, testing, and deployment

Expert guidance for writing secure, reliable, and performant Claude Code hooks - validates design decisions, enforces best practices, and prevents common pitfalls

Use when creating, improving, or troubleshooting Claude Code subagents. Expert guidance on agent design, system prompts, tool access, model selection, and best practices for building specialized AI assistants.

Use when creating OpenCode agents - provides markdown format with YAML frontmatter, mode/tools/permission configuration, and best practices for specialized AI assistants

Use when creating or improving Claude Code agents. Expert guidance on agent file structure, frontmatter, persona definition, tool access, model selection, and validation against schema.

Enforce evidence-based completion verification by requiring fresh execution of verification commands and confirmation of output before making any success claims, ensuring work is genuinely complete rather than assumed complete. Use this skill when about to claim that work is complete, finished, or done, when about to state that tests are passing or a test suite succeeds, when preparing to commit changes to version control, when about to create pull requests or merge requests, when claiming that a bug has been fixed or resolved, when stating that build processes succeed or compile without errors, when reporting that linting, formatting, or code quality checks pass, when delegating work to agents and receiving success reports that need independent verification, when moving from one task to the next in a multi-step implementation, when about to use words like "should work", "probably works", "seems to", "looks correct", or other qualifying language that implies uncertainty, when feeling satisfied with work and r

Use when creating GitHub Copilot instructions - provides repository-wide and path-specific formats, applyTo patterns, excludeAgent options, and natural language markdown style

Use when working with Anthropic Claude Agent SDK. Provides architecture guidance, implementation patterns, best practices, and common pitfalls.

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

Use when working with AWS Strands Agents SDK or Amazon Bedrock AgentCore platform for building AI agents. Provides architecture guidance, implementation patterns, deployment strategies, observability, quality evaluations, multi-agent orchestration, and MCP server integration.

Designs comprehensive testing strategies for any codebase. Use when adding tests, improving coverage, setting up testing infrastructure, or when asked about testing approaches.

Instructs AI agents on documentation standards for design docs, folder READMEs, source code interfaces, and test cases

Use the Claude Code CLI to consult Claude and delegate coding tasks for prototyping, debugging, and code review. Supports multi-turn sessions via SESSION_ID. Optimized for low-token, file/line-based handoff.

Post-process LaTeX project prose to improve readability through varied sentence and paragraph lengths. Removes filler phrases and unnecessary transitions while preserving all citations and semantic meaning.

Expert guidance for fine-tuning LLMs with Axolotl - YAML configs, 100+ models, LoRA/QLoRA, DPO/KTO/ORPO/GRPO, multimodal support

Implements and trains LLMs using Lightning AI's LitGPT with 20+ pretrained architectures (Llama, Gemma, Phi, Qwen, Mistral). Use when need clean model implementations, educational understanding of architectures, or production fine-tuning with LoRA/QLoRA. Single-file implementations, no abstraction layers.

High-performance RLHF framework with Ray+vLLM acceleration. Use for PPO, GRPO, RLOO, DPO training of large models (7B-70B+). Built on Ray, vLLM, ZeRO-3. 2× faster than DeepSpeedChat with distributed architecture and GPU resource sharing.

This skill should be used when the user asks to "initialize auto-memory", "create CLAUDE.md", "set up project memory", or runs the /auto-memory:init command. Analyzes codebase structure and generates CLAUDE.md files using the exact template format with AUTO-MANAGED markers.

Data framework for building LLM applications with RAG. Specializes in document ingestion (300+ connectors), indexing, and querying. Features vector indices, query engines, agents, and multi-modal support. Use for document Q&A, chatbots, knowledge retrieval, or building RAG pipelines. Best for data-centric LLM applications.

Native multi-agent orchestration using Claude Code's Agent Tool. Use when multiple independent tasks can run with different domain expertise or when comprehensive analysis requires multiple perspectives.