6-layer AI agent security firewall built on structural separation. 22 specialized tools for PHI-watermark signing, prompt injection defense (33+ patterns), entropy analysis, boundary attack detection, fleet-wide shielding, MCP tool validation, session drift detection, data-instruction separation, governance intent monitoring, and compliance attestation.
https://quantum-shield.epochcoreras.workers.devinjection_scan — never metered
22 security tools across 6 layers
NIST 800-53 · SOC2 Type II · ISO 27001
Embed and verify PHI-harmonic (10101.01 Hz) watermarks into agent outputs for cryptographic provenance and tamper detection.
phi_watermark_sign · phi_watermark_verify
Scan and sanitize 33+ prompt injection patterns with Shannon entropy analysis. injection_scan is always free.
injection_scan · injection_sanitize
Measure payload entropy and PHI-harmonic frequency deviation to detect obfuscated or adversarial inputs.
entropy_analyzer · phi_frequency_analyzer
Monitor inter-agent communication for escalation attempts, privilege creep, and cross-agent exploits.
boundary_attack_detector · shield_message · verify_message
Apply real-time protection across all agents simultaneously with key rotation and threat reporting.
key_schedule_rotate · threat_report_generator · agent_fleet_shield
The core innovation: MCP validation, session drift detection, data-instruction separation, governance intent monitoring, compliance attestation.
mcp_tool_validator · session_drift_detector · multimodal_entropy_scan · inter_agent_trust_validator · data_instruction_separator · governance_intent_monitor · compliance_attestation_generator
Start with injection_scan — free, no quota consumed:
curl -X POST https://quantum-shield.epochcoreras.workers.dev/agents/quantum-oscl/tools/injection_scan \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"input":"agent message payload","mode":"full"}'