⚠️ Beta  ·  MIT License  ·  SportMind Suite

AI agent bots
for Telegram.

Deploy SportMind-powered AI agents on Telegram in minutes. Pre-built prompts, configs, and five use cases — backed by SportMind's calibrated sports intelligence layer.

View on GitHub → How it works
4
Prompt files
5
Use cases
0
Paid dependencies
MIT
Licensed

How it works

Telegram Managed Bots —
the real flow.

Telegram's Managed Bots feature lets a manager bot create personal AI agents on behalf of users — natively, with a single tap. Each agent gets its own username and chat. Its intelligence is defined by your SportMind prompt.

01
User opens @LobsterClawBot
A Telegram Managed Bots manager. User taps "Create a New Bot".
02
Telegram shows a native confirmation
"LobsterClawBot would like to create and manage a chatbot on your behalf." Bot name and username are auto-suggested.
03
User taps Create
Child bot is live: "AI Agent #007 is ready. Its behaviour is defined by LobsterClawBot."
04
User messages the bot
Manager bot routes to your LLM, loaded with a SportMind system prompt from this kit's prompts/ directory.
05
Agent responds with structured SportMind intelligence
Pre-match signals, fan token analysis, sentiment, and macro events — drawn from SportMind's full 651-file intelligence layer across 42 sports.

Use cases

Five independently forkable bots.

Each use case ships with a complete config and a worked example. Clone only what you need.

Fan Token
Fan Token Trading Bot
Fan token signals, FTP PATH_2 supply event intelligence, CHI/TVS sentiment. Canonical example: $AFC Arsenal.
config/fan-token-trading-bot.yamlexamples/fan-token-trading-bot.md
Match Intelligence
Match Intelligence Bot
Pre-match briefings for supporter groups. SMS, AFS, rivalry multipliers, and match conditions for any club.
config/match-intelligence-bot.yaml
Calibration
Calibration Bot
Collect prediction records from your community. Feeds directly into SportMind's calibration base. First 10 contributors earn Founding Calibrator status.
config/calibration-bot.yaml
Sentiment
Sentiment Monitor
Passive or on-demand CHI and TVS sentiment snapshots for any fan token community. Divergence signals included.
config/sentiment-monitor.yaml
Macro
Macro Event Interpreter
MiCA, SEC/CFTC, Chiliz ecosystem events, and crypto cycle context interpreted through the SportMind macro layer.
config/macro-event-interpreter.yaml

Prompts

Intelligence built in.

Every prompt is pre-loaded with the SportMind metrics it depends on. Load as the system message in any OpenAI-compatible LLM call.

pre-match-signal.md
Full SMS, AFS, PS, and FTP PATH_2 reasoning chain. Produces structured JSON signal output with direction, confidence, modifiers, and flags.
SMSFTP PATH_2
sentiment-monitor.md
CHI and TVS band definitions, divergence signals, and four-quadrant interpretation logic for fan token community health.
CHITVS
price-movement-explainer.md
Seven-layer cause attribution chain. Checks FTP PATH_2 first, works through every modifier layer before assigning primary and secondary drivers.
All modifiers
macro-event-interpreter.md
MiCA, SEC/CFTC, Chiliz ecosystem, crypto cycle, and US first-mover context. Classifies events by impact direction, magnitude, and timeframe.
Macro layer

Quick start

Two paths. One kit.

Use @LobsterClawBot and be live in minutes. Or build your own manager bot with full control.

Path A — fastest
@LobsterClawBot
1Open @LobsterClawBot on Telegram
2Tap Create a New Bot — Telegram handles bot creation natively
3Load a prompt from prompts/ as your agent's system message
4Your SportMind agent is live — zero server setup required
Path B — full control
Self-hosted manager bot
1Create your manager bot via @BotFather, enable Managed Bots
2Clone the kit. Open config/bot-api-bootstrap.yaml
3Add your LLM key and set system_prompt_file to a prompt
4Deploy with python-telegram-bot, grammY, telegraf, or aiogram

Reference implementation

SMI — built on this kit.

SportMind's own Telegram agent — SMI — is the reference implementation of this kit. It monitors public sources, applies SportMind's reasoning frameworks, and delivers structured weekly briefings. SMI is what this kit makes possible. Fork this repository to build your own.

Governance

Community reviewed.
Founder not in the chain.

Pull requests are reviewed and merged by Trusted Reviewers — drawn from Founding Calibrators in the core SportMind library. The founder may contribute PRs like any other contributor.

Contributor
Anyone who opens a pull request. No prior approval needed. New use cases, prompt refinements, framework adapters.
No requirement
Trusted Reviewer
Review rights on pull requests. Nominated by existing Trusted Reviewers. 2 approvals required for prompts and configs.
Founding Calibrator
Maintainer
Can merge pull requests once review thresholds are met. Governance changes require 3 Maintainer approvals.
90 days + 10 reviews
Read GOVERNANCE.md →

The suite

Part of a larger system.

Each component is independently usable and MIT licensed.
Together they cover the full stack — from raw intelligence to live agent interfaces.

SportMind core library
651 skill files · 129 calibration records · 96% accuracy
The intelligence layer. 651 calibrated files across 42 sports. Every other component builds on this.
sportmind.dev → GitHub
Telegram AI Bot Starter Kit
4 prompts · 5 use cases · MIT licensed
Deploy SportMind-powered AI agents on Telegram via Managed Bots. Pre-built prompts, configs, and governance.
View page → GitHub
Fan Token Agentic Wallet Kit
4 layers · 5 examples · MIT licensed
Chiliz Chain integration. Fan token on-chain mechanics, FTP PATH_2 supply events, and wallet-layer intelligence.
View page → GitHub
Intelligence Agent
5 source categories · 3 delivery options · MIT licensed
Monitors public sources. Classifies signals. Delivers structured briefings. You decide what enters the library.
View page → GitHub

Free to use.
Free to build on.

MIT licensed. Any LLM. Any Bot API framework.
The kit is ready.

View on GitHub → Contribute