Attack Library

412 attack techniques across OWASP LLM Top 10, OWASP Agentic Top 10, MITRE ATLAS, NIST AI 100-2, AVID.

48 of 412 shown
AttackCategoryStandardsSophisticationEffectivenessLanguagesCorporaLast verified
DAN familyLLM-01 Direct
OWASP LLM-01
Medium
12%
en43 days ago
Role-play overrideLLM-01 Direct
OWASP LLM-01
Medium
18%
en, hi63 days ago
Crescendo multi-turnLLM-01 Direct
OWASP LLM-01MITRE T0051
High
34%
en33 days ago
Skeleton KeyLLM-01 Direct
OWASP LLM-01
High
28%
en23 days ago
Many-shot jailbreakLLM-01 Direct
OWASP LLM-01
High
31%
en23 days ago
Translation-based jailbreakLLM-01 Direct
OWASP LLM-01
Medium
22%
en, hi, ta33 days ago
Code-switching attack (Indic)LLM-01 Direct
OWASP LLM-01AVID
High
41%
hi-en, ta-en, te-en51 day ago
Devanagari-script encoding (Indic)LLM-01 Direct
OWASP LLM-01
High
38%
hi, mr42 days ago
ArtPrompt ASCII-artLLM-01 Direct
OWASP LLM-01
Medium
14%
en13 days ago
Base64 / ROT13 / Hex encodingLLM-01 Direct
OWASP LLM-01
Low
8%
en13 days ago
GCG suffix attackLLM-01 Direct
OWASP LLM-01
Frontier
26%
en23 days ago
AutoDANLLM-01 Direct
OWASP LLM-01
Frontier
24%
en13 days ago
Document-embedded instructionsLLM-01 Indirect
OWASP LLM-01MITRE T0051
High
62%
en, hi3today
Tool-response manipulationLLM-01 Indirect
OWASP LLM-01
High
47%
en23 days ago
Image OCR injectionLLM-01 Indirect
OWASP LLM-01Multimodal
High
38%
en23 days ago
Multi-document chained injectionLLM-01 Indirect
OWASP LLM-01
Frontier
51%
en13 days ago
HTML-style hidden instructionsLLM-01 Indirect
OWASP LLM-01
Medium
29%
en13 days ago
Search result poisoningLLM-01 Indirect
OWASP LLM-01
High
36%
en13 days ago
PII extraction probingLLM-02
OWASP LLM-02DPDP
Medium
19%
en, hi43 days ago
Aadhaar/PAN/IFSC extraction (Indic)LLM-02
OWASP LLM-02DPDP
High
27%
en, hi3yesterday
Training data extractionLLM-02
OWASP LLM-02
Frontier
14%
en13 days ago
Membership inferenceLLM-02
OWASP LLM-02MITRE
Frontier
9%
en13 days ago
System-prompt-mediated PII leakageLLM-02
OWASP LLM-02
Medium
22%
en13 days ago
Tool-argument manipulationLLM-06
OWASP LLM-06Agentic
High
33%
en23 days ago
Tool-sequence manipulationLLM-06
OWASP LLM-06Agentic
High
28%
en23 days ago
Unauthorized tool accessLLM-06
OWASP LLM-06
Medium
21%
en13 days ago
Action confusion attacksLLM-06
OWASP LLM-06
Medium
18%
en13 days ago
Direct system-prompt extractionLLM-07
OWASP LLM-07
Medium
24%
en13 days ago
Indirect extraction via output couplingLLM-07
OWASP LLM-07
High
31%
en13 days ago
Role-play extractionLLM-07
OWASP LLM-07
Medium
27%
en, hi13 days ago
Memory poisoningAgentic
OWASP Agentic-01
High
36%
en23 days ago
Goal hijackingAgentic
OWASP Agentic-02
High
32%
en13 days ago
Inter-agent trust violationAgentic
OWASP Agentic-03
Frontier
28%
en13 days ago
Sub-agent manipulationAgentic
OWASP Agentic-04
High
25%
en13 days ago
Sandbox escapeAgentic
OWASP Agentic-05
Frontier
14%
en13 days ago
Tool-result tampering simulationAgentic
OWASP Agentic
High
22%
en13 days ago
MCP Inspector RCE pattern (CVE-2025-49596)MCP
CVE-2025-49596
Frontier
78%
en1today
Cross-tenant leak patternMCP
MCPOWASP LLM-08
High
41%
en13 days ago
MCP server impersonationMCP
MCP
High
36%
en13 days ago
Tool-poisoning attacksMCP
MCP
High
44%
en13 days ago
MCP discovery-time attacksMCP
MCP
Medium
22%
en13 days ago
Image-text injectionMultimodal
OWASP LLM-01Multimodal
High
39%
en23 days ago
Audio adversarial inputsMultimodal
MITREMultimodal
Frontier
18%
en, hi13 days ago
Document-based attacksMultimodal
OWASP LLM-01
High
42%
en23 days ago
Cross-modal image+textMultimodal
OWASP LLM-01Multimodal
Frontier
36%
en13 days ago
Model artifact backdoor scanningSupply Chain
OWASP LLM-03
Frontier
11%
en13 days ago
Hugging Face model risk patternsSupply Chain
OWASP LLM-03
High
27%
en13 days ago
Dependency-injection attacksSupply Chain
OWASP LLM-03
Medium
21%
en13 days ago