Executive risk overview

LLM AppSec platform · 8 applications · live portfolio

Portfolio risk score

—

loading…

Critical findings

—

across 0 applications

Compliance posture

—

OWASP · SOC2 · MITRE · NIST

Production gate

—

Recent scans — llmguard runs from shared SQLite

0 runs

No scans yet — start one from the CI/CD app at /scan.

Application selector

asthera-ai/gatewayARS 7.4 · HighDeploy blockedCritical

Attack surface exposure

Prompt injection
88
Data exfiltration
72
Model evasion
55
Supply chain
30
Insecure output
80
Excessive agency
75

Compliance framework risk

OWASP LLM Top 10
72%
SOC 2 Type II
61%
MITRE ATLAS
55%
NIST AI RMF
70%

Vulnerability exposure

2

Critical

5

High

4

Medium

Sources — Wiz: 3 findings · Dependencies: 4 CVEs

Data exploitation risk

PII exposure
85
Tenant data leakage
90
Model output exploit
80
Audit log gaps
60

Endpoint exposure

POST /v1/chathigh

LLM inference · unfiltered output

POST /v1/tools/invokecritical

Wildcard tool scope

GET /v1/contextmedium

Redis context read · no TTL check

POST /v1/embedlow

Qdrant write · tenant-scoped

OWASP LLM Top 10 — this app

LLM01LLM02LLM03LLM04LLM05LLM06LLM07LLM08LLM09LLM10

OWASP LLM Top 10 — cross-application heatmap

gateway
cce
agents
analytics
ingestion
ui
rag-chatbot
llm-agent
LLM01
✓
✓
✗
~
✓
✓
✗
✗
LLM02
✗
✓
✗
~
✓
~
✗
~
LLM03
✓
—
—
—
~
—
—
✗
LLM04
~
✓
✗
✓
✓
✓
~
✓
LLM05
✓
✓
~
~
~
✓
✗
✗
LLM06
✗
✓
✗
✓
✓
✓
✗
✗
LLM07
✓
✓
✗
✓
—
✓
✗
✗
LLM08
~
✓
~
~
~
—
~
—
LLM09
✓
~
~
✓
—
✓
✗
✓
LLM10
~
✓
✗
✓
✓
✓
✗
~

Portfolio-level risk dimensions

Attack surface by category

Risk score trend — last 6 releases

Security Node Graph · Code → Weakness → Control Gap

gatewayARS 7.4 · BLOCKED
Highest-risk service · illustrative flow
CodePOST /chatrag_chat.py:42
WeaknessLLM01Prompt Injection
CRITICAL GAPNo input filteron /chat endpoint
CodePOST /chatrag_chat.py:42
WeaknessCVE-2024-35195requests 2.31 · CVSS 9.1
CRITICAL GAPNo auth on /admin/admin/logs unprotected

5-layer ontology: Code · Exposure · Weakness · Vulnerability · Control Gap

View full security graph→