Build autonomous
agents with WASP
A powerful, minimal framework for creating autonomous agents using WebAssembly and Rust.
Everything you need
Powerful tools designed to make agent development intuitive and efficient
Hot Reloading
Instant feedback with our hot-reloading development environment for rapid iteration
WASM Performance
High-performance execution powered by WebAssembly and Rust's zero-cost abstractions
Interactive Console
Test your agents with rich commands, conversation history, and detailed metrics
Multiple LLM Support
Seamless integration with UOMI and third-party language models via a unified API
Debugging Tools
Comprehensive tools for debugging, performance monitoring, and error tracing
Version Control
Track agent evolution and development with built-in versioning and lifecycle management
Elegant by design
Write clean, concise code that's both powerful and easy to understand. WASP abstracts the complexity while giving you full control.
Type-safe Development
Leverage Rust's powerful type system
WebAssembly Optimization
Compiled to efficient, portable WASM
Intuitive Developer API
Clean, ergonomic interface for developers
use serde::{Deserialize, Serialize};
use utils::log;
mod utils;
#[derive(Serialize, Deserialize, Debug)]
struct Message {
role: String,
content: String,
}
fn parse_input(input: &str) -> Vec<Message> {
// Parse the input string into a JSON Value
let parsed: Vec<Message> = serde_json::from_str(input)
.expect("Failed to parse input JSON");
parsed
}
fn system_message(content: String) -> Message {
Message {
role: "system".to_string(),
content,
}
}
fn process_messages(
system_message: Message,
mut messages: Vec<Message>) -> Vec<Message> {
messages.insert(0, system_message);
messages
}
#[no_mangle]
pub extern "C" fn run() {
log("Start agent execution!");
//get the input
let input = utils::read_input();
let input = String::from_utf8_lossy(&input);
let input = parse_input(&input);
//create a system message
let system_message = system_message("You are a UOMI Agent!".to_string());
let modified_messages = process_messages(system_message, input);
let ai_input= serde_json::json!({
"messages": modified_messages,
}).to_string();
let ai_input_bytes = ai_input.as_bytes().to_vec();
let ai_response = utils::call_ai_service(1, ai_input_bytes);
let ai_response_str = String::from_utf8(ai_response.clone()).unwrap();
let ai_response_content = extract_ai_response_content(ai_response_str);
let ai_response_content_bytes = ai_response_content.as_bytes().to_vec();
// save output
utils::save_output(&ai_response_content_bytes);
}