Struct ToolOutputSanitizer 

pub struct ToolOutputSanitizer { /* private fields */ }

Removes malicious content from tool responses before they reach the agent.

This is the most critical component — tools can return content from external sources (web pages, emails, database results) that may contain prompt injection attacks targeting the agent.

Implementations

impl ToolOutputSanitizer

pub fn new() -> Self

Create a new output sanitizer with default patterns and settings.

Default maximum output length is 50,000 characters. HTML stripping is enabled by default.

pub fn with_strip_html(self, strip: bool) -> Self

Set whether HTML/script tags should be stripped.

pub fn with_max_output_length(self, max: usize) -> Self

Set the maximum output length (in characters).

pub fn sanitize( &self, output: &str, _tool_context: &ToolContext, ) -> SanitizeResult

Sanitize tool output by removing malicious content.

Returns a SanitizeResult containing the cleaned text, detections, and the worst severity found.

Trait Implementations

impl Debug for ToolOutputSanitizer

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ToolOutputSanitizer

fn default() -> Self

Returns the “default value” for a type.