Your Scraper Didn’t Fail. It Just Started Lying. So I Built Molt

说明与用途

Molt通过将抓取结果与基线比较来检测和修复隐藏的抓取错误，诊断漂移，生成针对性修复提示，并在应用修复前要求人工批准。

问题

Scrapers can succeed technically while producing incorrect data, leading to silent failures that go unnoticed by standard monitoring checks.

结果

Molt enabled automated detection of data drift, diagnosis of affected fields, generation of precise heal prompts, and a human‑in‑the‑loop approval gate, closing the loop from anomaly detection to verified data recovery.

工作流

• Run scraper and snapshot output
• Compare snapshot to baseline to detect drift
• Diagnose affected fields and quantify drift
• Generate a targeted heal prompt describing changes
• Human reviews the heal prompt and approves or rejects
• Apply heal via Bright Data CLI (bdata scraper heal)
• Re-run scraper to verify data recovery

证据

The job was still successful. HTTP was still 200. The row count was still 60. And download_count = 0 is a perfectly plausible number.

来源

Molt treats scraper breakage more like an SRE incident than a failed request.

来源