| name | prompt-engineering |
| description | Prompt engineering knowledge base — technique taxonomy with decision tree, prompt template patterns and formatting conventions, OWASP LLM Top 10 security checklist, eval frameworks and testing guide, context engineering, structured output contracts, multi-agent orchestration patterns, cost optimization. Use when designing prompts, reviewing prompt quality, building AI features, creating AI assets, or auditing LLM security. |
| user-invocable | false |
Prompt Engineering
Comprehensive prompt engineering knowledge base. Provides actionable patterns, checklists, and guides for designing, securing, evaluating, and optimizing LLM prompts and agent systems.
When to Use
- Designing or reviewing prompt templates (system, developer, user prompts)
- Building tool calling schemas and structured output contracts
- Evaluating prompt quality — accuracy, safety, cost, latency
- Auditing LLM security against OWASP LLM Top 10
- Designing multi-agent orchestration and handoff protocols
- Optimizing prompt cost and latency
- Creating or reviewing Claude Code AI assets (rules, workflows, skills — all are prompts)
- Setting up prompt versioning and observability
When NOT to Use
- Implementing backend/frontend code (use
Agent(software-engineer) + stack-specific role)
- Infrastructure and deployment (use
Agent(devops-engineer))
- Writing code tests (use
Agent(qa-engineer) + testing-procedures skill)
- Content writing (use
Agent(content-writer))
- Context pipeline design, memory engineering, agent harness, RAG architecture, multi-agent orchestration, production AI checklists → use
context-engineering skill
Key Concepts
Prompt as System
A prompt is not a string — it is a system composed of:
- Instruction hierarchy: System prompt > Developer prompt > User prompt > Retrieved content
- Context assembly: What enters the context window, in what order, with what priority
- Output contract: Schema, format, constraints, error handling, fallback behavior
- Tool interface: Available tools, their schemas, permissions, composition patterns
- Guard rails: Safety filters, refusal policies, output validators
- Versioning: Immutable versions, deployment tags, audit trail
Core Principles
- Eval-first: Define how to measure before changing anything
- Simplest technique: Zero-shot → few-shot → CoT → chaining. Escalate only when simpler fails
- Explicit over implicit: Spell out constraints, output format, edge cases. Never assume the model "knows"
- Separation of concerns: Instructions vs data vs examples — always delimited
- Grounding: Prefer citations and verifiable data over unanchored claims
- Least privilege: Minimal tool permissions per agent. HITL for high-impact actions
- Cost awareness: Every token costs money and time. Compress, cache, route
Resource Files
| File | Contents |
|---|
technique-guide.md | Full technique taxonomy with decision tree, examples, and anti-patterns |
prompt-template-patterns.md | Delimiter conventions, system prompt structure, few-shot formatting, CoT triggers, output schema patterns |
security-checklist.md | OWASP LLM Top 10 mapped to prompt-level mitigations with checklist |
eval-and-testing-guide.md | Eval frameworks, grader types, dataset curation, A/B testing, regression gates |
Integration
- Follows rules:
Agent(prompt-engineer) (prompt system architecture, security, eval-first quality)
- Used by workflows:
/ai-assets (all assets are prompts), /feature-dev (AI features), /code-review (prompt quality review)
- Companion skills:
context-engineering skill (context pipeline design, memory engineering, agent harness, RAG architecture, multi-agent orchestration, production checklists), asset-validation skill (AI asset format validation), code-review skill (review checklists)
- Collaborates with roles:
Agent(software-engineer) (prompt integration), Agent(qa-engineer) (prompt regression tests), Agent(product-manager) (success metrics)
