Monitors public sources. Classifies signals. Applies the Library Rule. Delivers structured briefings with context and reasoning. You decide what enters the library.
Agent reasoning chain
Every run follows the same pipeline. Every decision is logged. The briefing you receive shows exactly what was found, why it was classified, and what to do about it.
The governing principle
Before any signal reaches the briefing it must pass the SportMind Library Rule:
"Will this intelligence still be true and useful in six months?"
If yes — it is enduring reasoning that belongs in the library. If no — it is expiring data that belongs in the briefing body only.
The agent applies this test automatically. Expiring signals appear as context. Only enduring reasoning frameworks are routed to the library.
Source monitoring
Every source is publicly accessible, RSS or API-based, and ToS-compliant. No paywalled content. No scraping. No ToS violations.
Briefing format
Tier 1 signals always include all seven fields. Every claim is traceable to its source. Every recommended change names a specific SportMind file and section.
Delivery
Choose the method that fits your workflow. GitHub Issue is the default — auditable, searchable, zero extra infrastructure.
Reference implementation
SMI is SportMind's own Intelligence Agent — the reference implementation of this repository. It delivers weekly briefings to a private Telegram group, fires immediate alerts for Tier 1 signals, and has been running since May 2026. Fork this repository to build your own.
The suite
Each component is independently usable and MIT licensed.
Together they cover the full stack — from raw intelligence to live agent interfaces.
MIT licensed. Works with any LLM.
Zero autonomous commits. You stay in control.