Batch Wave SVG Generation with the JAD Runner and API

Fetch the schema before building payloads

The GET endpoint tells you the exact option fields and ranges. Build your variant config against this so you never send a non-existent field.

curl -H "Authorization: Bearer YOUR_KEY" \
  https://jadapps.com/api/v1/tools/svg-wave-divider

# → { ..., "input": { "isGenerative": true, "options": [
#       {"name":"waveAmplitude","min":5,"max":120,"default":40}, ... ] },
#     "execution": { "runnerMode": "engine" }, "minTier": "free" }

Why POSTing to the cloud /run returns 400

The cloud /run is upload-free and never executes — it returns 400 with the runner endpoint. This is expected; switch your POST target to the localhost runner.

curl -X POST -H "Authorization: Bearer YOUR_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"options":{"waveAmplitude":40,"waveColor":"#6366f1"}}' \
  https://jadapps.com/api/v1/tools/svg-wave-divider/run

# HTTP 400
# { "error": "Runner required",
#   "runnerEndpoint": "http://127.0.0.1:9789/v1/tools/svg-wave-divider/run" }

Loop a colour matrix against the local runner

Read the schema, then iterate your section/colour matrix in Node, dispatching each payload to the paired runner and writing the SVG to public/.

const variants = [
  { name: 'hero',    waveColor: '#ffffff', waveAmplitude: 40, waveFrequency: 3 },
  { name: 'pricing', waveColor: '#0f172a', waveAmplitude: 25, waveFrequency: 4 },
  { name: 'cta',     waveColor: '#6366f1', waveAmplitude: 60, waveFrequency: 2, waveFlipped: true },
];
for (const v of variants) {
  const r = await fetch('http://127.0.0.1:9789/v1/tools/svg-wave-divider/run', {
    method: 'POST', headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({ options: v }),
  });
  fs.writeFileSync(`public/wave-${v.name}.svg`, await r.text());
}

Per-breakpoint amplitude, not per-breakpoint width

There is no width option — the viewBox is fixed at 1440 and stretches via preserveAspectRatio. For mobile, vary amplitude (a narrow viewport compresses the curve) rather than expecting a width parameter.

// Desktop and a punchier mobile variant of the same colour:
const desktop = { name: 'hero-lg', waveColor: '#fff', waveAmplitude: 40 };
const mobile  = { name: 'hero-sm', waveColor: '#fff', waveAmplitude: 70 };
// Serve via <picture><source media="(max-width:640px)" srcset="/hero-sm.svg"></picture>

Deterministic shapes without a seed

Identical numeric options always yield an identical path — the sine is deterministic, so the same amplitude/frequency/height across colour variants produces sibling shapes automatically. No seed field is needed or available.

const shape = { waveAmplitude: 40, waveFrequency: 3, waveHeight: 80 };
['#fff','#0f172a','#6366f1'].forEach((c, i) =>
  dispatch({ ...shape, waveColor: c, _name: `wave-${i}` })
);
// All three share the exact same curve; only the fill differs.

By design the cloud /api/v1/tools/svg-wave-divider/run is upload-free and never executes — it returns HTTP 400 with runnerEndpoint pointing at http://127.0.0.1:9789/.... Switch your POST target to the local runner. This is the intended flow, not an error in your request.

Both the GET schema and POST run endpoints call validateApiKey and reject missing or invalid keys with an auth error before doing anything else. Org-issued keys also carry per-tool scopes — a key not scoped for this tool is denied. Mint and scope a key first.

validateSvgOptions enforces the schema min/max: waveAmplitude must be ≤120, waveFrequency ≤10, waveHeight ≤240. An out-of-range number throws a descriptive error instead of producing degenerate output. Clamp your matrix to the documented ranges.

If a numeric field receives a value that isn't finite (e.g. "abc" or NaN), validation throws must be a number. Cast your config values to numbers before dispatching. Omitting a field is fine — the engine default applies.

These fields are not in the schema, so the runner ignores them — they do not error, but they do nothing. Don't build your config types around them; the only knobs are the five documented options.

Both endpoints enforce a per-key hourly rate limit and return 429 with X-RateLimit-* headers and a resetAt. For large matrices, batch within the limit or pace your loop. Execution itself is local, but each schema/dispatch call still counts against the cloud rate limit where applicable.

The response is the SVG (engine mode), with no computed contrast metadata. To validate text-over-wave contrast, compute it yourself from waveColor against your overlay colour using a WCAG contrast formula in your own script.

The browser download is always wave-divider.svg. In automation you control the filename when you write the response to disk — use your variant name (wave-hero.svg). Don't rely on the tool to disambiguate names for you.