Concept

Prompt Engineering

Designing instructions, examples and context so LLMs behave reliably — formally studied and applied across my agents and specs.

  • #ai
  • #llms

Prompt engineering is the practice of shaping model inputs — instructions, examples, structure and context — to get accurate, consistent output. In agent systems it extends to tool descriptions, system prompts and context management.

How I use it

  • mcp-deep-dive — treats MCP prompts (parametrized templates) as a first-class capability, which most public examples skip.
  • local-rag-chatbot — prompt-driven multi-query expansion and a grading prompt that decides whether context is sufficient.
  • Akashi — context management through a per-step hook with last-N windowing and summarization of older turns by a cheaper model.
  • Spec-Driven Development — the specs I write are, in effect, structured prompts for coding agents.

Credentials and courses

Generative AI: Prompt Engineering in the IBM Generative AI Engineering Professional Certificate certificate.

Related