ScrapeForensics

Incident response for self-healing web scrapers. When a target site changes and a scraper breaks, ScrapeForensics detects it, figures out why, triggers Bright Data's self-healing, verifies the fix actually worked, and keeps a forensic record of the whole incident.

BREAK → INVESTIGATE → EXPLAIN → HEAL → VERIFY → REMEMBER

Built for WeMakeDevs Into the Scrape-Verse (Bright Data, Aug 17–23 2026).

Structure

Folder	Role	Owns
collector/	Bright Data + Collector infrastructure	Create/run/heal/approve wrappers, the incident store, verification
diagnosis/	Scraper Doctor + AI diagnosis	INVESTIGATE + EXPLAIN — what broke, why, in plain language
chaos-testing/	Chaos testing + verification	BREAK — controlled mutations + the Reliability Score

Each folder has its own README with setup and API details. diagnosis and chaos-testing both build on top of collector's exported functions (runCollector, healCollector, verifyHeal, store.*) rather than talking to Bright Data directly — keeps one place responsible for the CLI integration.

Setup

```bash npm install -g @brightdata/cli bdata login ```

Claim hackathon credits: sign up at brightdata.com, enter promo code `wemakedevs` (lowercase) in the billing section.

Then see each folder's README for its own install/run steps.

Judging alignment

• Use of Scraper Studio — collector/ wraps bdata scraper create/run/heal/approve directly
• Reliability and self-healing — the entire premise; chaos-testing/ turns it into a measured score
• Technical excellence / presentation — the collector/demo.ts script runs the full break→heal→verify→record loop end to end for the live demo