Zero External Pip Dependencies
Sub-3ms Inline Latency
Port 8099 Proxy
Neutralize Prompt Injections. Scrub Secrets In Real Time.
Autonomous inline security proxy and threat radar for AI agents. Intercepts LLM input prompts and outbound completions to eliminate jailbreaks, delimiter collisions, credential exfiltration, and PII leaks before transmission.
AEGIS HEXAGONAL THREAT RADAR • 360° ACTIVE SWEEP
Total Prompts Inspected
0
Attacks Neutralized
0
Secrets & PII Redacted
0
Clean Pipeline Rate
100%
SSE LIVE RADAR ACTIVE (:8098)
Intercepted Prompts: 0
Live Interception Radar & Threat Telemetry Feed
Broadcasts prompt injection intercepts, secret scrub events, and latency in real time via Server-Sent Events.
Streaming Interception Log Feed
Listening on /api/events/stream...
Waiting for Prompt Ingestion Stream
Send requests through proxy port
:8099 or test prompts below. Threats and sanitizations stream here in real time with zero page reloads.
Real-Time Threat Score Velocity
Block Threshold: 0.70
Interactive Threat Sandbox & Sanitizer
Test real-world prompt injection payloads, system prompt exfiltration probes, and secret exposure vectors.
Input Prompt Payload
Real-Time Threat Diagnostics
CLEAN
Decision Outcome:
PERMITTED ✅
Threat Severity Score:
0.00
Inspection Latency:
1.85ms
Secrets / PII Scrubbed:
0
Triggered Heuristics:
- No active threat signatures detected.
Sanitized Wire Output:
Enterprise Compliance & Custom Heuristics
Automated SOC-2 Type II audit exports, HIPAA de-identification manifests, and custom AST de-obfuscation.
SOC-2 Type II Logs
Immutable hash-chained JSON logs of every prompt evaluation and decision.
AST De-obfuscation
Decodes recursive base64, zlib, unicode homoglyphs, and zero-width spaces.
Custom Regex Rules
Inject proprietary data boundaries and custom organization secret markers.