Prompt Engineering

What it does

Provides a comprehensive catalog of prompt design techniques, structural templates (XML-style tags), agentic rules, and best practices to control output shape, handle ambiguity, and reduce hallucinations. Includes checklists for audits, model-specific guidance, and a prompting techniques catalog for common use cases.

When to use it

Use when creating or reviewing prompts for chatbots, agentic systems, research evaluations, or production pipelines — especially for tasks requiring structured outputs, tool use, or long-context grounding. Triggers on requests to "improve this prompt", "write a system prompt", "audit my prompt", or when building agent toolchains.

What's included

• Scripts: none in this skill (has_scripts=false).
• References: yes — a references/ folder with deeper guides and checklists.
• Instructions: explains XML-like sectioning (<context>, <instructions>, <examples>), output-specs, uncertainty handling, long-context grounding, tool usage rules, and self-checks for high-risk outputs.

Compatible agents

Intended for LLM-based agents and tooling such as Claude Code, GPT-family agents, Cursor, Codex-style CLIs, and any agent supporting structured prompts and tool integrations.