ai-prompting

name	ai-prompting
description	Quick tips and templates for effective prompt engineering - few-shot examples, chain-of-thought patterns, constraint specification, output formatting
short_desc	quick prompting templates and tips
keywords	["few-shot examples","constraint specification","output formatting","prompt template","prompting tips","prompt patterns","prompt engineering","improve this prompt","write a prompt"]
model	haiku

AI Prompting (Haiku)

Purpose: Quick tips and templates for effective prompt engineering - few-shot examples, chain-of-thought patterns, constraint specification, output formatting.

Model: Haiku 4.5 (fast, practical prompt improvement)

When to Invoke Autonomously:

Use this skill when:

1. Prompt Not Working Well: "LLM output is inconsistent/poor quality"
2. Need Examples: "How to structure few-shot prompt for [task]?"
3. Output Formatting: "LLM not following desired format"
4. Constraint Specification: "How to enforce constraints in prompt?"
5. Chain-of-Thought: "Need step-by-step reasoning for [complex task]"

DO NOT invoke for:

• Complex prompt engineering (use AI LLM Expert agent)
• Production prompt systems (use AI LLM Expert agent)
• Already-working prompts

Usage

/ai-prompting improve-prompt [current prompt] [issue]
/ai-prompting few-shot [task description]
/ai-prompting chain-of-thought [reasoning task]
/ai-prompting output-format [desired structure]

Quick Prompt Patterns

For detailed examples of each pattern, see examples/prompt-patterns.md.

Common Patterns

1. Few-Shot Examples: Show 2-5 input→output examples (15-30% accuracy improvement)
2. Chain-of-Thought: "Think step-by-step" for reasoning tasks
3. Output Format Specification: Explicit structure with examples
4. Constraint Enforcement: Explicit boundaries, "DO NOT" statements
5. Role-Based Prompting: "You are a [role] specializing in [domain]" (20-40% improvement)

Common Issues & Fixes

• Inconsistent Output → Explicit format + examples
• Over-Verbose → Length constraints
• Hallucinations → Constrain to provided context only
• Ignoring Instructions → Repeat constraints, add examples

Quick Workflow Reference

Before implementing: Search for proven patterns

.claude/scripts/kg-search search "prompt-engineering" --type concepts

For deep research: Ask user "Use hybrid_search to research [prompting techniques]"

Development env: Python 3.12, Weaviate:8081, Ollama:11435, venv: source claude_mcp_servers/.venv/bin/activate

Integration with Knowledge Graph

After prompt improvement:

1. Document successful pattern in knowledge/prompts/[use-case]-patterns.md
2. Tag with use case and effectiveness

Supporting Files

• Prompt Patterns: See examples/prompt-patterns.md for detailed examples
• Template: Use template.md for structured prompt improvements

Success Metrics

• ✅ Prompt output improves after applying pattern
• ✅ Tips are actionable and immediate
• ✅ Common issues resolved