Special Tasks from Ton Station

Partner: Ton Station (TS)

Category: Development of agents and analytical tools based on SIDUS AI

Goal: Creation of applied solutions that enhance the Ton Station ecosystem using Sidus AI tools (plugins, LLM agents, integrations, and AI analytics).

General Requirements

  • All projects must be implemented using the SIDUS AI Framework.
  • Code must be published in a private GitHub repository.
  • Individual participants and teams are allowed.
  • Interface: web/telegram/console — at the developers' discretion.
  • The final deliverable must include an MVP, a description of the logic, example prompts (if necessary), and interaction flow.

Prize Pool

$2,500
1st place
$1,200
2nd place
$800
3rd place
$500

List of Tracks from TON STATION

AG-1. Agent for Analyzing Social Media Activity

Task

  • Develop a SIDUS AI agent or plugin that analyzes activity on social networks (Telegram, X/Twitter, YouTube, Reddit, etc.) based on specified keywords or users;

Objectives

  • Identify mentions of specified keywords (e.g., Ton Station, TON, SIDUS, AI Hackathon);
  • Track engagement and sentiment (positive/negative);
  • Identify the most active channels and influential users mentioning the given topic;

Functionality

  • Setting source parameters (Telegram channels, X/Twitter, YouTube, Reddit, etc.);
  • Setting key phrases (e.g., "tonstation", "TS", "TON", etc.);
  • Generating an aggregated report:
  • Number of mentions;
  • Activity index;
  • Popular authors and posts;
  • Change in engagement over time;
  • Automatic segmentation by channels and users;
  • Ability to set filters (time range, language, platform);
  • Analytics visualization (charts, dashboard);
  • Use of built-in LLM insights - a brief "AI summary" on audience sentiment;

AG-2. Personal Channel "Mr. Soon" (MR. Soon)

Task

  • Bring the $MRSOON character to life for managing social media;

Objectives

  • Publish content (news, insights, jokes, reviews);
  • Teach the mascot to interact with the audience;
  • Develop a "personality" and communication style through training;

Functionality

  • The agent integrates with a Telegram channel (or other media channel);
  • Uses an LLM profile for "MR Soon";
  • Character: charismatic, slightly edgy, with a sense of humor;
  • Automatically publishes posts 1-2 times per day, after editor review;
  • Analyzes audience reaction and adjusts its style (via a feedback loop);
  • Stores a content database and engagement statistics;
  • Creation of a web-based "control panel" interface for MR Soon;

AG-3. AI Advisor Agent for Ton Station Efficiency

Task

  • Create an agent that analyzes the application "like a human" and, after a period of learning, begins to use it independently, performing only the most profitable actions for the user who launched it;

Objectives

  • Create a self-learning system that provides the user with the best action strategy within the Ton Station ecosystem;
  • Automate decision-making in Ton Station to maximize profit and minimize user error;

Functionality

  • Building a model of the user's "digital behavior" within the ecosystem;
  • Control of PC/Mobile;
  • Generation of daily insights from the application;
  • Automatic adaptation of recommendations based on user progress and engagement;
  • Ability to connect to a Telegram bot for daily notifications and interactive reports;
  • "Gamification AI" module: the agent assesses the user's contribution and suggests missions and challenges;
  • Integration with other SIDUS AI agents (e.g., analytics or the MR Soon content bot);
  • Capability for self-learning based on user behavior (the agent becomes smarter over time);

Submission Format

  • Private GitHub Repository;
  • README.md file with description and setup instructions;
  • Demo video up to 3 minutes long;
  • (Optional) Telegram bot or web interface to demonstrate the agent's functionality.