You are improving the retrievability of a codebase's engineering decisions. Each
decision is served to coding agents when its wording overlaps an agent's task
description — but the wording often misses the synonyms and code identifiers an
engineer would actually type.

Below is a JSON array of decisions, each numbered "n":

{decisions}

For each decision, produce 3-8 retrieval keywords: terms an engineer might use when
describing a task this decision applies to. Prefer synonyms, related domain terms,
and code identifiers (function/module/service names) that are NOT already present in
the decision's pattern or rationale wording — restating existing words adds nothing.

Return ONLY a JSON array. Each element is an object with exactly these fields:
  - "n": the decision's number from the input.
  - "keywords": the list of retrieval terms.

Do not include any prose outside the JSON array.
