Differentiate JSON-shape illustrations from real JSON request bodies

[jstirnaman]

Background

In several places in our docs, API request body samples contain a long Flux/SQL script as a JSON string value. Authors have historically formatted these with \ line continuations and unescaped inner " for readability:

{
  "script": "from(bucket: \"example-bucket\")\
              |> range(start: -task.every)\
              |> filter(fn: (r) => r._measurement == \"mem\" ...)",
  "language": "flux"
}

These are not valid JSON, but they're also not literal — they're shell/curl-style examples where bash line continuation joins the lines before the JSON parser sees the body. The \ and unquoted-inner-" are docs-readability conventions, not literal request bodies.

Today this means the blocks fall back to a text fence (no syntax highlighting, no parse-lint coverage). PR #7170 made that explicit during the JSON lint cleanup.

Problem

Readers can't tell from a text fence whether a block is:

• A real JSON request body (escape it, paste it, send it)
• A JSON-shape illustration (you'd never send this verbatim — you'd build it from the prose)

That ambiguity is fine for a couple of examples but recurs across process-data/*, api-guide/*, and the v1 Kapacitor API docs.

Proposal

Standardize on two paired blocks for any request body that embeds a multi-line script as a JSON string field. Pattern, applied as a reference in content/shared/influxdb-v2/process-data/get-started.md "Full example task with invokable script":

The Flux for the script field:

```js ... readable Flux block ... ```

The full request body, with the script encoded as a JSON string (inner quotes escaped as \", line breaks as \n):

```json ... valid JSON with \n-escaped script ... ```

Benefits:

• Both blocks lint clean (Flux block can stay js-highlighted; JSON block actually parses).
• Reader sees the script in its native form.
• The encoding step is taught explicitly — no hidden shell magic.

For places where the shell wrapping itself is the lesson (env-var interpolation, heredoc usage), use tabs-wrapper with "Request body" / "curl command" tabs instead.

Files to convert

Known instances (others may exist):

• content/shared/influxdb-v2/process-data/get-started.md — "Define and use scripts" section (aggregate-intervals block, ~line 147)
• content/shared/influxdb-v2/process-data/get-started.md — "Full example task with invokable script" (aggregate-intervals-and-export block, ~line 220) — converted in fix(lint): resolve JSON blocking errors (39 of 53) #7170 as the reference
• content/influxdb/cloud/api-guide/api-invokable-scripts/_index.md — filter-and-group block (~line 126)
• Audit remaining \``text` blocks where the body is JSON-shaped to find others

Out of scope

• Telegraf-generated content under content/telegraf/v1/{input,output,processor}-plugins/ — those are synced from influxdata/telegraf and need upstream PRs.
• Real RECORD BATCH / msgpack output blocks — those genuinely aren't JSON and stay text.

Reference

PR #7170 fixed JSON parse failures across content/ and surfaced this pattern. See review thread for the trade-off discussion that led to this proposal.