Senthex

A security platform for AI agent chains that detects prompt injections hidden in incoming content that propagate across chained agents without triggering guardrails, and provides a verifiable execution log for audits. It is for developers, IT teams and B2B organizations building and operating agent workflows. It is delivered as a SaaS proxy that intercepts requests via a base_url swap and scans inputs and outputs across 26 shields while generating a hash-chained, signed and timestamped audit trail.

Key features

• Red teaming for AI agent chains
• One-week assessment with report and debrief
• Verifiable execution log
• SHA-256 hash chain integrity
• Merkle-committed event set
• RFC 3161 external timestamping
• Ed25519 + ML-DSA-65 origin signatures
• 26 shields per call scanning
• Input and output scanning
• Streaming pass-through proxy
• EU-hosted infrastructure
• Base_url swap integration
• Offline log verifier