WebGuardian Platform Architecture
Platform Thesis
Observe → Understand → Repair → Validate → Deploy → Learn
WebGuardian AI Operational Layers
EXPERIENCE LAYERControl Center & Operator Cockpit
DashboardIncident War RoomData Explorer
INTELLIGENCE LAYERDecision Graph & Diagnostics Engine
Drift EngineLangGraph AgentAgent Memory
EXECUTION LAYERManaged Proxy Fabric & Sandbox Sandbox
Bright Data CollectorsProxy RotationValidation Sandbox
Chronological Self-Healing Data Lifecycle
1. Target Website DOM
Original DOM v1 shifts to redesigned layout v2.
TELEMETRY: Price container class '.price' disappears; tag becomes '[data-testid=price]'
2. Bright Data Collector
Runs unblocking proxies and headless scraper scripts.
TELEMETRY: Monitored via unique API triggers and snapshot IDs
3. Scraper Observatory
Audits data drops and structural validation mismatches.
TELEMETRY: Translates 0-row extraction drops into pipeline failures
4. Drift Engine
Classifies anomalies: Schema Drift, DOM Drift, or Runtime drops.
TELEMETRY: Blocks selector repair loops on runtime connection timeouts
5. LangGraph Repair Agent
Runs state machine node loops (Triage, Analysis, intent, Plan).
TELEMETRY: Recovers field extraction contract details to design solutions
6. Validation Sandbox
Runs candidate selectors on new DOM and ranks outcomes.
TELEMETRY: Formula: 30% Semantic + 30% Coverage + 20% Schema + 10% Layout + 10% Confidence
7. Studio Version Deploy
Publishes configuration updates (vN) to Scraper Studio.
TELEMETRY: Deprecates active v1 tags and triggers immediate pipeline recovery
8. Recovered Pipeline
Delivers complete dataset rows. Downtime: <35s.
TELEMETRY: Logs version audit traces and preserves rollback triggers