OWASP Top 10 for LLMs
612Signatures
98%Coverage
12 min agoLast Sync
- LLM01: Prompt Injection
- LLM02: Sensitive Information Disclosure
- LLM05: Improper Output Handling
- LLM06: Excessive Agency
Total Signatures
4,812
86 this week
Active Behavioral Models
37
3 this week
Threat Sources
5
All syncing
Custom Threat Packs
4
1 installed this month
Coverage Score
94.2%
1.6% vs last sync
Last Sync
12 min ago
Next in 48 min
Threat Intelligence Sources
Signature and behavioral model coverage across every framework this deployment defends against
Active
OWASP Agentic AI Risks
348Signatures
91%Coverage
12 min agoLast Sync
- ASI01: Agent Goal Manipulation
- ASI04: Excessive Tool Permissions
- ASI07: Multi-Agent Collusion
- ASI09: Memory Poisoning
MITRE ATT&CK
2,184Signatures
96%Coverage
9 min agoLast Sync
- T1059: Command and Scripting Interpreter
- T1499: Endpoint Denial of Service
- T1552: Unsecured Credentials
- T1071: Application Layer Protocol
MITRE ATLAS
1,204Signatures
93%Coverage
9 min agoLast Sync
- AML.T0051: LLM Prompt Injection
- AML.T0048: Goal Manipulation
- AML.T0043: Craft Adversarial Data
- AML.T0024: Exfiltration via ML Inference API
Custom Enterprise Threats
464Signatures
87%Coverage
38 min agoLast Sync
- CUST-004: Payroll Data Exfil Pattern
- CUST-011: Internal Tool Impersonation
- CUST-017: Support Macro Abuse
Signature Distribution by Source
4,812Signatures
- OWASP Top 10 for LLMs13% (612)
- OWASP Agentic AI Risks7% (348)
- MITRE ATT&CK45% (2,184)
- MITRE ATLAS25% (1,204)
- Custom Enterprise Threats10% (464)
Recent Signature Updates
18 new signatures added for LLM01: Prompt Injection
ASI09 behavioral model retrained on latest telemetry
MITRE ATT&CK synced to v15.1, 42 techniques updated
AML.T0051 coverage expanded with 6 new variants
CUST-004 exfil pattern updated after incident review
Acme Corp Internal Threat Pack updated to v1.0.3
Signature Library
| Signature ID | Name | Source | Category | Severity | Status | Last Updated |
|---|---|---|---|---|---|---|
| LLM01-0142 | Instruction Override via Roleplay Persona | OWASP LLM Top 10 | Prompt Injection | Critical | Active | Sep 9, 2026 |
| LLM02-0087 | System Prompt Extraction Pattern | OWASP LLM Top 10 | Sensitive Info Disclosure | High | Active | Sep 8, 2026 |
| LLM05-0033 | Unsanitized Markdown Rendering Exploit | OWASP LLM Top 10 | Improper Output Handling | Medium | Active | Sep 3, 2026 |
| ASI01-0021 | Agent Objective Redirection via Tool Output | OWASP Agentic AI | Goal Manipulation | Critical | Active | Sep 9, 2026 |
| ASI04-0014 | Over-Scoped Tool Grant Detection | OWASP Agentic AI | Excessive Permissions | High | Active | Sep 6, 2026 |
| ASI09-0009 | Long-Term Memory Poisoning via Feedback Loop | OWASP Agentic AI | Memory Poisoning | High | Draft | Sep 2, 2026 |
| T1059-3312 | Scripting Interpreter Invocation from Agent Shell | MITRE ATT&CK | Execution | High | Active | Sep 7, 2026 |
| T1552-1187 | Unsecured Credential Read via Tool Call | MITRE ATT&CK | Credential Access | Critical | Active | Sep 9, 2026 |
| T1499-0876 | API Rate Flood Consistent with Resource Exhaustion | MITRE ATT&CK | Impact | Medium | Active | Sep 1, 2026 |
| T1071-0245 | Covert C2 over Application Layer Protocol | MITRE ATT&CK | Command and Control | High | Disabled | Aug 24, 2026 |
| AML.T0051-018 | Indirect Prompt Injection via Retrieved Document | MITRE ATLAS | Initial Access | Critical | Active | Sep 9, 2026 |
| AML.T0048-006 | Agent Goal Drift Behavioral Signature | MITRE ATLAS | Goal Manipulation | Medium | Active | Sep 5, 2026 |
| AML.T0024-011 | Model Inversion via Repeated Inference Queries | MITRE ATLAS | Exfiltration | High | Active | Sep 4, 2026 |
| CUST-004-002 | Payroll Database Exfil Phrase Pattern | Custom Enterprise | Data Exfiltration | Critical | Active | Sep 8, 2026 |
| CUST-011-005 | Internal Helpdesk Tool Impersonation | Custom Enterprise | Tool Misuse | Medium | Draft | Aug 30, 2026 |
Showing 15 of 15 signatures
Enterprise Threat Packs
Installable signature bundles for enterprise-specific and vertical-specific threats
Financial Services Threat Pack
InstalledSignatures tuned for wire-transfer fraud, PCI data handling and regulated advisory agents.
Healthcare & PHI Guard Pack
InstalledHIPAA-aligned detections for clinical assistant agents handling patient records.
Acme Corp Internal Threat Pack
InstalledEnterprise-authored signatures for internal tooling, helpdesk macros and payroll workflows.
DevOps Agent Hardening Pack
AvailableCoverage for infrastructure-as-code agents, credential stores and CI/CD tool misuse.
Legal & Contract Review Pack
AvailableDetections for privileged-document leakage and contract-term manipulation attempts.