Struct CompactContextTool 

pub struct CompactContextTool;

Server-side tool for manually triggering conversation context compression.

Unlike automatic compression (which triggers based on token thresholds), this tool lets the model or user proactively compress history at natural task boundaries. It bypasses the exposure gate and always forces a compression cycle.

Trait Implementations

impl Tool for CompactContextTool

fn name(&self) -> &str

fn description(&self) -> &str

Human-readable tool description for LLM.

fn parameters_schema(&self) -> Value

JSON Schema for tool parameters.

fn execute<'life0, 'async_trait>( &'life0 self, args: Value, ) -> Pin<Box<dyn Future<Output = Result<ToolResult, ToolError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Execute the tool with given arguments.

fn execute_with_context<'life0, 'life1, 'async_trait>( &'life0 self, args: Value, _ctx: ToolExecutionContext<'life1>, ) -> Pin<Box<dyn Future<Output = Result<ToolResult, ToolError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Execute the tool with a streaming-capable context.

fn mutability(&self) -> ToolMutability

Declares whether this tool is read-only or mutating for orchestration and parallel scheduling decisions. Defaults to mutating to stay conservative.

fn call_mutability(&self, _args: &Value) -> ToolMutability

Args-aware mutability hook. Defaults to the static mutability declaration.

fn concurrency_safe(&self) -> bool

Declares whether this tool can safely run in parallel with other read-only tools. Defaults to false so tools remain serialized unless they opt in explicitly.

fn call_concurrency_safe(&self, _args: &Value) -> bool

Args-aware parallel-safety hook. Defaults to the static declaration.

fn to_schema(&self) -> ToolSchema

Convert tool to LLM-compatible schema.