Case study

A real autonomous agent, running in production.

xbot is an autonomous Twitter/X application built using RustyAI's agent-core and cognition crates.

4

scheduled posts / day

5

content categories

2

crates: agent-core, cognition

24/7

long-lived runtime behaviour

Pipeline

Scheduler → agent → cognition → content → safety → publish.

  1. Scheduler

    4 scheduled posts per day

  2. xbot Agent

    lifecycle, state, long-lived execution

  3. Cognition

    beliefs about topics and recent output

  4. Content Generation

    5 content categories, AI-generated

  5. Safety Filter

    content screened before publication

  6. Twitter / X

    publish

xbot.rs
use rustyai::prelude::*;
#[tokio::main]
async fn main() -> Result<()> {
    let xbot = AgentBuilder::new("xbot")
        .with_belief("audience_prefers_technical")
        .with_goal("post_daily")
        .schedule(Cron::daily_at(&["08:00", "12:00", "16:00", "20:00"]))
        .build();
    Runtime::builder()
        .supervisor(RestartPolicy::exponential_backoff())
        .spawn(xbot)
        .run()
        .await
}

Why it matters

xbot exercises the parts of an agent framework that only surface over time: lifecycle management across restarts, belief state that has to stay coherent between runs, scheduled autonomous execution, and a safety boundary between generation and publication.

Build your own long-lived agent.

Start from the same primitives xbot uses: agent-core for lifecycle, cognition for beliefs and goals.

Get StartedRead the architecture