Update confidence levels prompt injection detection to reduce false positive rates (#6390)

Signed-off-by: Dorien Koelemeijer <dkoelemeijer@squareup.com>
This commit is contained in:
dorien-koelemeijer
2026-01-08 15:51:53 +10:00
committed by GitHub
parent 9dc548ee2f
commit 3d415fc1ea
4 changed files with 13 additions and 13 deletions
+2 -2
View File
@@ -85,7 +85,7 @@ impl PromptInjectionScanner {
pub fn get_threshold_from_config(&self) -> f32 {
Config::global()
.get_param::<f64>("SECURITY_PROMPT_THRESHOLD")
.unwrap_or(0.7) as f32
.unwrap_or(0.8) as f32
}
pub async fn analyze_tool_call_with_context(
@@ -283,7 +283,7 @@ mod tests {
let scanner = PromptInjectionScanner::new();
let result = scanner.analyze_text("rm -rf /").await.unwrap();
assert!(result.confidence > 0.9);
assert!(result.confidence >= 0.75); // High risk level = 0.75 confidence
assert!(!result.pattern_matches.is_empty());
}