One-shot prompting evaluates what a model or agent can produce from a single request. The initial prompt must carry the objective, constraints and expected result because the user does not refine the work through follow-up turns.
The method can make comparisons easier when systems receive the same prompt, but it does not represent every real workflow. Iterative feedback often improves results and may be part of normal application development.
ELI5
One-shot prompting means giving an AI helper one set of instructions and judging what it produces without corrections. The helper gets one chance to understand and complete the job.
For example, two coding agents may receive the same request for a playable game and submit their first complete builds before anyone gives feedback.
