Provides deterministic scanning and detection of prompt injection, jailbreak, data-exfiltration, and system-prompt leaks, with EIP-191 signed attestations and 0G Storage anchoring for verifiable safety reports.
Enables testing AI safety classifier robustness against query decomposition, obfuscation, and multi-agent attacks. Provides tools for full evaluation pipelines, query previews, and status checks.
Provides real-time, privacy-preserving analysis of LLM interactions to detect problematic behaviors like medical advice, dangerous file operations, physics speculation, and unsupported claims. Recommends safety interventions and builds a taxonomy of LLM limitations through crowdsourced evidence collection.
An independent review layer for AI-generated text: a 10-lens gate run by a separate model over output it did not write, returning the flagged line and the reason for each check rather than a single opaque verdict. A check that cannot complete reports unknown instead of a pass; lens_health needs no API key.
Enables security researchers to evaluate AI system defenses against prompt injection attacks through a comprehensive set of test vectors and analysis tools.
