Prompt Engineering

Prompt engineering is the craft of wording an instruction to a language model so it produces the desired output reliably. Where context engineering decides what the model sees, prompt engineering decides how that instruction is phrased — role and system prompts, few-shot examples, chain-of-thought elicitation, output-format constraints, and the failure modes that come from ambiguous or over-stuffed prompts.

Some of the output is not the prompt’s to fix: the vocabulary an aligned model reaches for by default is a property of its training, and LLM overused words covers what a banned-word list buys and where it stops working. The same holds above the word, for sentence patterns and reply openers: mitigating verbal tics covers which prompt instructions have been measured to lower their rate.

Wording an instruction to produce a voice is its own problem with its own measurements: pastiche covers what a style prompt can and cannot reach, and rules versus examples compares showing the model prose against describing it.

For a concrete instance — how the rules and the page request are worded for an agent that then acts on them — see the section’s running example, Claude Code: writing a page for this wiki.

This page is a stub. In the meantime:

Backward Claude Code LLM Overused Words Forward