List your company

System Prompt Exfiltration on Cline Agent Environment

100Pseudo-RecallToolLeak
-4235077Sep 6, 2025

Evaluation results

MethodLinks
ToolLeakBackend LLM=sonnet-42025.09100
ToolLeakBackend LLM=sonnet-4.52025.09100
ToolLeakBackend LLM=gemini-2.5-pro2025.09100
ToolLeakBackend LLM=gpt-52025.09100
ToolLeakBackend LLM=grok-42025.0981
NaiveBackend LLM=grok-42025.0962
Completion-RememberBackend LLM=gemini-2.5-pro2025.0924
Remember-the-StartBackend LLM=grok-42025.093
NaiveBackend LLM=sonnet-42025.090
NaiveBackend LLM=sonnet-4.52025.090
NaiveBackend LLM=gemini-2.5-pro2025.090
NaiveBackend LLM=gpt-52025.090
IgnoreBackend LLM=grok-42025.090
IgnoreBackend LLM=sonnet-42025.090
IgnoreBackend LLM=sonnet-4.52025.090
IgnoreBackend LLM=gemini-2.5-pro2025.090
IgnoreBackend LLM=gpt-52025.090
CompletionBackend LLM=grok-42025.090
CompletionBackend LLM=sonnet-42025.090
CompletionBackend LLM=sonnet-4.52025.090
CompletionBackend LLM=gemini-2.5-pro2025.090
CompletionBackend LLM=gpt-52025.090
Ignore-CompletionBackend LLM=grok-42025.090
Ignore-CompletionBackend LLM=sonnet-42025.090
Ignore-CompletionBackend LLM=sonnet-4.52025.090
Ignore-CompletionBackend LLM=gemini-2.5-pro2025.090
Ignore-CompletionBackend LLM=gpt-52025.090
Remember-the-StartBackend LLM=sonnet-42025.090
Remember-the-StartBackend LLM=sonnet-4.52025.090
Remember-the-StartBackend LLM=gemini-2.5-pro2025.090
Remember-the-StartBackend LLM=gpt-52025.090
Ignore-RememberBackend LLM=grok-42025.090
Ignore-RememberBackend LLM=sonnet-42025.090
Ignore-RememberBackend LLM=sonnet-4.52025.090
Ignore-RememberBackend LLM=gemini-2.5-pro2025.090
Ignore-RememberBackend LLM=gpt-52025.090
Completion-RememberBackend LLM=grok-42025.090
Completion-RememberBackend LLM=sonnet-42025.090
Completion-RememberBackend LLM=sonnet-4.52025.090
Completion-RememberBackend LLM=gpt-52025.090
Ignore-Completion-RememberBackend LLM=grok-42025.090
Ignore-Completion-RememberBackend LLM=sonnet-42025.090
Ignore-Completion-RememberBackend LLM=sonnet-4.52025.090
Ignore-Completion-RememberBackend LLM=gemini-2.5-pro2025.090
Ignore-Completion-RememberBackend LLM=gpt-52025.090
PLeakBackend LLM=grok-42025.090
PLeakBackend LLM=sonnet-42025.090
PLeakBackend LLM=sonnet-4.52025.090
PLeakBackend LLM=gemini-2.5-pro2025.090
PLeakBackend LLM=gpt-52025.090