v148 · on-device AI

Safety Harness

12 edge-case tests: API availability, invalid performance values, empty and very-long input, model-unavailable recovery, format/length/type combinations, concurrent summarisation, and graceful fallback paths.

Checking Summarizer API…

12Total

0Pass

0Fail

0Warn

Ready. Press "Run all" or individual test buttons.

Test suite

// Defensive Summarizer usage
async function safeSummarize(text, pref = 'balanced') {
  if (!window.Summarizer) return { error: 'not-supported' };
  if (!text?.trim()) return { error: 'empty-input' };

  const valid = ['fast', 'balanced', 'quality'];
  if (!valid.includes(pref)) pref = 'balanced'; // clamp

  const cap = await Summarizer.capabilities();
  if (cap.available === 'no') return { error: 'model-unavailable' };

  try {
    const s = await Summarizer.create({ performancePreference: pref });
    return { result: await s.summarize(text) };
  } catch (e) {
    return { error: 'inference-error', message: e.message };
  }
}

implementation reference

Need the exact API surface, compatibility boundaries, errors, lifecycle, and source links? Read the matching gendn reference ↗

see also

implementation reference