Skip to main content

Prompts

In the Basics chapter, we learned how to get a simple prompt. In this section, we’ll explore in more detail how deal with resources provided by the MCP server.

Getting a Prompt​

To get a prompt, use client.prompts().get(). It requires the prompt name and optional arguments.

use neva::prelude::*;

#[tokio::main]
async fn main() -> Result<(), Error> {
    let mut client = Client::new()
        .with_options(|opt| opt
            .with_stdio(
                "cargo", 
                ["run", "--manifest-path", "./neva-mcp-server/Cargo.toml"]));

    client.connect().await?;

    let args = ("lang", "Rust");
    let prompt = client.prompts().get("hello_world_code", args).await?;

    println!("{prompt.descr:?}: {prompt.messages:?}");

    client.disconnect().await
}

Passing Arguments​

If a prompt requires a single parameter, pass a tuple containing the parameter name and its value:

let args = ("lang", "Rust");
let prompt = client.prompts().get("hello_world_code", args).await?;

If a prompt requires multiple parameters, pass them as an array, Vec, or HashMap:

let args = [
    ("lang", "Rust"),
    ("topic", "Hello World function"),
];
let prompt = client.prompts().get("write_code", args).await?;

If a prompt is parameterless, pass the unit type ():

let prompt = client.prompts().get("rust_hello_world", ()).await?;

Listing Prompts​

list(cursor) asks for one page of the server's prompts, and list_all() walks every page:

for prompt in client.prompts().list_all().await? {
    println!("{}: {:?}", prompt.name, prompt.descr);
}

A prompt is the user's to pick, not the model's to call. To open a conversation with one, neva::svir::prompt_messages turns it into the messages a model is sent.

Learn By Example​

Here you may find the full example