What is worldcup2026-prediction-skill?

This repository is a single system prompt file designed to turn any compatible AI language model into a 2026 FIFA World Cup prediction engine. The project is called worldcup2026-prediction-skill, and its main deliverable is a file named skill.md that you paste or load as the system message when calling an AI API. The prompt bundles data about all 48 teams competing in the 2026 World Cup directly inside itself. That includes key players, recent form, injury concerns, coaching staff, and historical matchup notes. The model is instructed to reference only this embedded data when making predictions, and it is not allowed to invent match histories or injury reports that are not written there. Predictions are scored using four weighted dimensions: recent form counts for 40 percent of the result, squad strength for 30 percent, historical head-to-head records for 15 percent, and situational factors such as home continent advantage and squad age for the remaining 15 percent. A built-in ceiling prevents the model from assigning more than 85 percent win probability to any team, which acknowledges that upsets are a genuine part of football. The output is locked to a strict JSON schema, meaning every prediction comes back in the same structured format regardless of which AI model is doing the reasoning. That makes it straightforward to wire the results into a website or app without extra parsing. The prompt also includes a daily briefing section at the end where match-day updates such as lineup changes can be inserted without rewriting the rest of the file. Usage requires any API that accepts the OpenAI chat format. The README shows working code samples in Python, Node.js, and plain curl. No training, no vector database, and no backend server are needed beyond a standard API key.