Protection levels are templated guardrail strategies that help operators get to a sane starting posture quickly. Protection Level 3 is the recommended starting point.Documentation Index
Fetch the complete documentation index at: https://docs.protectorplus.cloudsine.tech/llms.txt
Use this file to discover all available pages before exploring further.
| Level | Default state | Notes |
|---|---|---|
| Level 0 — No Protection | All guardrails off | Default for new profiles. |
| Level 1 — Standard Protection | Standard guardrails available; none on by default | Operator opts in per guardrail. |
| Level 2 — Standard + ShieldPrompt Preset 1 | LLM + Vector on | Foundational coverage for prompt-injection. |
| Level 3 — Standard + ShieldPrompt Preset 2 (Recommended) | LLM + Vector + Content Moderation + System Prompt Protection on | Strong default for production. |
| Full Custom | Operator-defined mix | Choose any combination across Standard Protection and ShieldPrompt. |
Choosing a level
Customer-facing chatbot
Level 3 + tightened PII categories for your jurisdiction.
Internal RAG / knowledge agent
Level 3 + System Prompt Protection enabled in-line; consider Vector at High.
Agentic / multi-tool agent
Level 3 + Vector at High sensitivity, plus tight Keyword/Regex rules on tool-call surfaces.
Research / red-team sandbox
Level 1 or Level 2 in Monitor mode — observe without blocking.