⚡AgentSkills
🤖 AI Engineering · Prompt Engineering

Architect a maintainable system prompt

Organize identity, rules, tools, and output contracts into sections that evolve without breaking behavior.

foundation~30 minAI EngineersML EngineersLLM App Developers

Steps

  1. 1Section order: identity, capabilities, hard rules, task guidance, output format, examples
  2. 2State hard rules as short imperatives; explain rationale only when it changes behavior
  3. 3Put the output contract last and show one canonical example
  4. 4Version the prompt like code; change one concern per revision
  5. 5Keep a regression set of 15 prompts run after every edit
  6. 6Delete anything that survived three revisions without effect

Common Pitfalls

  • ▲Conflicting rules buried mid-prompt that silently override each other
  • ▲Examples contradicting the stated format

Success Signals

  • ✓Regression pass rate 100% on releases
  • ✓Prompt length trending down

Commands

Install with skills CLI
$ npx skills add aniruddhaadak80/skills --skill prompt-engineering-system-prompt-arch
Install globally
$ npx skills add aniruddhaadak80/skills --skill prompt-engineering-system-prompt-arch -g

Tags

#prompts#llm#ai-engineering#prompt-engineering

Related skills

Select few-shot examples that move accuracy (Classification tasks)intermediate

Choose and order examples covering edge cases so the model generalizes instead of copying surface patterns.

🤖 AI Engineering·~30m
Open →
Select few-shot examples that move accuracy (Extraction tasks)intermediate

Choose and order examples covering edge cases so the model generalizes instead of copying surface patterns.

🤖 AI Engineering·~30m
Open →
Guarantee structured outputs from LLMsintermediate

Get machine-parseable JSON reliably using schema-first prompting, constrained decoding, and repair loops.

🤖 AI Engineering·~35m
Open →