Frequently asked questions

What Cloak Harbor is, what it is not, and how the synthetic demo stays safe.

Is Cloak Harbor a people-search tool?
No. Cloak Harbor is a defensive Personal Exposure Defense control plane. It helps authorized operators reduce a consented principal's exposure. It is not a people-search engine, doxxing system, or public OSINT crawler.
Does the demo use real data?
No. The demo uses only a synthetic principal (Jordan Hale, synthetic-principal-001) and reserved/fictional identifiers. No real people are scraped or enriched in demo mode.
What is StealthScore?
StealthScore is a rubric-driven posture score that converts validated exposure findings into a defensible 0–100 risk number plus a projected remediation delta. In the demo, projected scores are labeled demo calibration.
Can the AI submit takedowns automatically?
Never. AI only drafts notices in dry-run mode. Every external action requires explicit human approval, and drafts are always labeled AI-drafted.
How is access controlled?
The demo environment is designed to sit behind firewall / IP allowlisting. State-changing API actions are gated server-side by PLATPHORM_API_KEY, which is never exposed client-side.
What is SigReduce?
SigReduce is the open-core foundation: the exposure object schema, evidence model, identity graph structure, risk taxonomy, scoring rubric and synthetic dataset. We do not open-source any personal-data scraper.
What happens if the AI Gateway is unavailable?
The system degrades gracefully to deterministic template drafts, clearly labeled as template fallback. It never fabricates model output.