🤖Sovereign AI Architectural CopilotGroq • GPT-OSS 120B

Explore Rullst's architecture. Local answers can reference public demo posts; optional cloud answers use your message.

🛡️ Rullst AI Firewall: ACTIVE

✨ Showcase Copilot

Welcome to the Sovereign Showcase AI Laboratory! I can answer questions about the 5 Web Paradigms, explain active SQLite posts, provide Rust code examples, or test adversarial inputs against the Prompt Injection Shield. What would you like to explore?

Check this box to use cloud AI. Send my message to Groq or the configured AI provider (optional).Unchecked = local replies. Avoid personal or confidential data. Privacy details

🛡️Prompt Injection Shield Arena (Interactive Pentesting)

Test adversarial prompts against the live Rullst AI Firewall. The defense-in-depth model neutralizes jailbreak attempts before they can manipulate system context.

✅ Valid Architectural Queries (Safe):

⚠️ Adversarial Jailbreak Attempts (Test Shield Defense):

// Rullst Defense-in-Depth Specification:
• Layer 1 (Heuristic Firewall): Token pattern inspection blocks known jailbreak payloads.
• Layer 2 (Grounding): RAG system prompt confines LLM knowledge to database records.
• Layer 3 (Output Sanitizer): HTML escaping prevents stored or reflected XSS injection.
• Layer 4 (MicroVM Isolation): Ephemeral Azure Container App runs rootless with no host mounts.
• Layer 5 (Scale-to-Zero Reset): Non-persistent memory; pristine defaults restore on wake.