langchain_rust::tools

Trait Tool

Source 
pub trait Tool: Send + Sync {
    // Required methods
    fn name(&self) -> String;
    fn description(&self) -> String;
    fn run<'life0, 'async_trait>(
        &'life0 self,
        input: Value,
    ) -> Pin<Box<dyn Future<Output = Result<String, Box<dyn Error>>> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait;

    // Provided methods
    fn parameters(&self) -> Value { ... }
    fn call<'life0, 'life1, 'async_trait>(
        &'life0 self,
        input: &'life1 str,
    ) -> Pin<Box<dyn Future<Output = Result<String, Box<dyn Error>>> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait,
             'life1: 'async_trait { ... }
    fn parse_input<'life0, 'life1, 'async_trait>(
        &'life0 self,
        input: &'life1 str,
    ) -> Pin<Box<dyn Future<Output = Value> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait,
             'life1: 'async_trait { ... }
}

Required Methods§

Source

fn name(&self) -> String

Returns the name of the tool.

Source

fn description(&self) -> String

Provides a description of what the tool does and when to use it.

Source

fn run<'life0, 'async_trait>( &'life0 self, input: Value, ) -> Pin<Box<dyn Future<Output = Result<String, Box<dyn Error>>> + Send + 'async_trait>>
where Self: 'async_trait, 'life0: 'async_trait,

Executes the core functionality of the tool.

Example implementation:

ⓘ
async fn run(&self, input: Value) -> Result<String, Box<dyn Error>> {
    let input_str = input.as_str().ok_or("Input should be a string")?;
    self.simple_search(input_str).await
}

Provided Methods§

Source

fn parameters(&self) -> Value

This are the parametters for OpenAi-like function call. You should return a jsnon like this one

{
    "type": "object",
    "properties": {
        "command": {
            "type": "string",
            "description": "The raw command you want executed"
                }
    },
    "required": ["command"]
}

If there s no implementation the defaul will be the self.description()
Source

fn call<'life0, 'life1, 'async_trait>( &'life0 self, input: &'life1 str, ) -> Pin<Box<dyn Future<Output = Result<String, Box<dyn Error>>> + Send + 'async_trait>>
where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Processes an input string and executes the tool’s functionality, returning a Result.

This function utilizes parse_input to parse the input and then calls run. Its used by the Agent

Source

fn parse_input<'life0, 'life1, 'async_trait>( &'life0 self, input: &'life1 str, ) -> Pin<Box<dyn Future<Output = Value> + Send + 'async_trait>>
where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Parses the input string, which could be a JSON value or a raw string, depending on the LLM model.

Implement this function to extract the parameters needed for your tool. If a simple string is sufficient, the default implementation can be used.

Implementors§