Textbox
Select All
Deselect All
Filter by category
adversarial
agent_manipulation
benign
chain_of_thought
code_execution
context_confusion
control
crescendo
direct_injection
edge_case
encoding
encoding_obfuscation
indirect_injection
instruction_override
jailbreak
many_shot
model_fingerprinting
multi_turn
output_manipulation
payload_injection
persona_replacement
prompt_extraction
prompt_injection
prompt_leak
rag_poisoning
response_manipulation
system_extraction
system_manipulation
token_injection
token_smuggling
training_extraction
t-SNE Space
Click a point to analyze it. Hover to preview text. Use scroll wheel to zoom.
Analysis Result
Full prompt
Select a prompt
Search dataset
Or analyze a custom prompt
Custom prompt
Analyze
GuardLLM
- Prompt Security Visualizer
Model:
Llama Prompt Guard 2 (86M)
by Meta | Dataset:
neuralchemy/Prompt-injection-dataset