prompt-engineering — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited prompt-engineering (Plugin) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
<p align="center"> <h1 align="center">Awesome Prompt Engineering Skill</h1> </p>
<p align="center"> <strong>The prompt engineering knowledge base that turns Claude Code into a prompt engineering expert — from Chain of Thought to production system prompt architecture.</strong> </p>
<p align="center"> <a href="#why-this-exists">Why</a> • <a href="#whats-inside">What's Inside</a> • <a href="#quick-start">Quick Start</a> • <a href="#usage">Usage</a> • <a href="#resources">Resources</a> </p>
<p align="center"> <a href="https://opensource.org/licenses/MIT"><img src="https://img.shields.io/badge/License-MIT-yellow.svg" alt="License: MIT"/></a> <img src="https://img.shields.io/badge/Claude%20Code-Skill-blueviolet" alt="Claude Code Skill"/> <img src="https://img.shields.io/badge/Techniques-58%2B-blue" alt="58+ Techniques"/> <img src="https://img.shields.io/badge/Reference%20Docs-7-green" alt="7 Reference Docs"/> </p>
Most prompt engineering guides teach you what to write. This skill teaches you how to architect prompts — the patterns behind production agentic systems that handle millions of requests.
We analyzed Claude Code's leaked system prompts and the Awesome Prompt Engineering research catalog into one actionable skill. The result goes beyond prompt content:
This isn't another "add 'Let's think step by step'" guide. It's the engineering discipline behind prompts that actually ship.
| # | Capability | What You Get |
|---|---|---|
| 1 | Core Prompting Techniques | CoT, ToT, ReAct, few-shot, constraint injection, distributed reinforcement |
| 2 | System Prompt Architecture | Modular section-builders, static/dynamic boundaries, prompt caching, context injection |
| 3 | Agent & Tool Integration | Agent specialization (5 types), tool-aware prompts, tiered permissions, Think tool |
| 4 | Prompt Optimization | APE, DSPy, EvoPrompt, compression, A/B testing workflows |
| 5 | Security & Robustness | Injection defense layers, instruction hierarchy, Constitutional AI, red teaming |
| 6 | Evaluation & Benchmarking | Assertion-based, model-graded, human eval, regression testing |
| 7 | Production Best Practices | Prompt-as-code, versioning, monitoring, context management, anti-patterns |
| File | Content |
|---|---|
techniques-catalog.md | 58+ prompting techniques organized by category with templates |
architecture-patterns.md | System prompt architecture derived from Claude Code analysis |
agent-patterns.md | Multi-agent delegation, specialization, concurrency patterns |
security-guide.md | Defense in depth, injection defense, red team methodology |
optimization-tools.md | APE, DSPy, Promptfoo, manual optimization workflows |
evaluation-frameworks.md | Benchmarks (SWE-bench, MMLU-Pro), testing tools, CI/CD |
production-checklist.md | Deployment checklist, monitoring, anti-patterns |
Register this repo as a marketplace, then install the skill — two commands:
/plugin marketplace add hardness1020/awesome-prompt-skill
/plugin install prompt-engineering@awesome-prompt-skillThat's it. Claude Code automatically activates the skill when you ask about prompt engineering.
# Copy the skill directly
cp -r skills/prompt-engineering /path/to/your/project/.claude/skills/"Help me design a system prompt for a customer support agent"
"How should I structure my multi-agent system's prompts?"
"Review my prompt for injection vulnerabilities"
"What's the best way to add few-shot examples?"
"Help me set up prompt testing with Promptfoo"
"How do I optimize my prompt's token usage?"| If you need to... | Capability |
|---|---|
| Write or improve a prompt | Core Techniques |
| Design a complex system prompt | Architecture |
| Build multi-agent or tool-using systems | Agent & Tool Integration |
| Systematically improve prompt performance | Optimization |
| Harden prompts against attacks | Security |
| Test and benchmark prompts | Evaluation |
| Deploy prompts to production | Production |
| Paper | Contribution |
|---|---|
| The Prompt Report | Survey of 58 text + 40 multimodal techniques |
| Chain of Thought Prompting | Step-by-step reasoning elicitation |
| Tree of Thoughts | Tree search over reasoning paths |
| ReAct | Interleaving reasoning with tool use |
| Constitutional AI | Self-alignment via behavioral principles |
| DSPy | Declarative prompt programming framework |
| Self-Consistency | Multiple reasoning paths, majority vote |
| APE | Automated prompt engineering via LLMs |
| Many-Shot ICL | Scaling in-context learning to 100s-1000s of examples |
| Graph of Thoughts | Arbitrary graph reasoning (62% quality improvement) |
| Tool | Purpose | Link |
|---|---|---|
| Promptfoo | CLI prompt testing & red-teaming | github.com/promptfoo/promptfoo |
| DSPy | Declarative prompt programming | github.com/stanfordnlp/dspy |
| LangChain | LLM application framework | github.com/langchain-ai/langchain |
| LangGPT | Structured prompt design | github.com/EmbraceAGI/LangGPT |
| Garak | LLM vulnerability scanner | github.com/NVIDIA/garak |
| Langfuse | LLM observability | github.com/langfuse/langfuse |
| DeepEval | RAG & agent evaluation | github.com/confident-ai/deepeval |
| CrewAI | Multi-agent orchestration | github.com/crewAIInc/crewAI |
| Resource | Type |
|---|---|
| Prompt Engineering Guide | Comprehensive guide (DAIR.AI, 55K+ stars) |
| Learn Prompting | Interactive learning platform |
| ChatGPT Prompt Engineering for Developers | Free course (DeepLearning.AI + OpenAI) |
| Stanford CS336 | Language Modeling from Scratch |
This skill was built by analyzing and consolidating knowledge from:
MIT License - Copyright 2026 Marcus Chang
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.