About worldcup-predictor
A skill that runs deterministic World Cup predictions using an audited offline data snapshot and bundled prediction scripts. It is intended for match and tournament analysis, with explicit rules for preserving source data, separating result types, and reporting uncertainty.
Key features
- 01Predictions from audited offline snapshots
- 02Match and tournament simulations
- 03Lottery reference list generation
- 04Snapshot validation and provenance checks
- 05Separate 90-minute and advancement results
Use cases
- 01Estimating a match from a prepared snapshot
- 02Simulating a tournament with a seed
- 03Generating a 3/1/0 lottery reference list
Install
Copy the command and run it in a terminal.
npx skills add https:/ / github.com/ qqyule/ worldcup-predictor-skill --skill worldcup-predictorAfter installing, type/worldcup-predictorin Claude Code to confirm it loaded.
Bundled files
References and scripts installed along with SKILL.md.
README.en.md.gitignoreLICENSEREADME.mdpackage.jsonagents/openai.yamlassets/official-sources.jsonassets/sample-data/synthetic-48-team.jsonlottery-issue.jsonworldcup-2026.jsoncore/betting.mjstournament.mjsindex.mjstypes.mjsutils.mjsversion.mjsmanifest.jsonmatch.mjsreferences/tournament-rules.mdmodel-methodology.mdofficial-data-sources.mddata-schema.mdlottery-rules.mdscripts/generate-lottery-slip.mjspredict-match.mjsaudit-input.mjssimulate-tournament.mjstests/skill.test.mjsDetails
From the project README世界杯预测机 / World Cup Predictor
Use this skill to run deterministic World Cup predictions from an offline audited snapshot. Users may call it 世界杯预测机 in Chinese prompts; the internal skill name remains worldcup-predictor. The bundled core/ is a portable snapshot of prediction-core v0.2.0; do not recreate probability formulas in prose or with an LLM.
Workflow
- Obtain a structured audited snapshot. Do not scrape or invent missing official facts.
- Read
references/data-schema.mdand validate source versions, one complete strength version, and completed match fields. - If official provenance is relevant, read
references/official-data-sources.md; use its source index only to audit snapshot lineage, not as live prediction input. - Run the relevant CLI from the skill directory:
node scripts/predict-match.mjs --data <snapshot> --home FRA --away BRA
node scripts/simulate-tournament.mjs --data <snapshot> --simulations 10000 --seed 2026
node scripts/generate-lottery-slip.mjs --issue <issue> --strategy balanced --budget 288
- Explain only the returned probabilities and audit metadata. State uncertainty and fallback status.
- Keep
90minResultandadvanceResultseparate in every report.
Non-Negotiable Rules
- Use
90minResultonly for 3/1/0 lists, group points, and 90-minute predictions. - Use
advanceResultonly for knockout progression and champion paths. - Preserve completed group scores and completed knockout advancing teams.
- Apply host advantage only when
venueCountryCodematches the team's country. - Treat
officialFacts, weather, and news as audit context only. - Ignore
llm_extractionadjustments. Apply onlymanual_reviewor versioneddeterministic_ruleadjustments. - Never claim guaranteed accuracy, returns, purchasing advice, or official endorsement.
Bundled Data
assets/sample-data/worldcup-2026.json: compact synthetic audited smoke-test snapshot.assets/sample-data/synthetic-48-team.json: synthetic 48-team snapshot with 73 completed matches.assets/official-sources.json: lightweight official source registry metadata only.- Samples are not official feeds and contain no licensed marks or crests. The source registry does not include official data, raw responses, media, or live feeds.
References
- Read
references/data-schema.mdbefore preparing or validating snapshots. - Read
references/official-data-sources.mdbefore assessing official source provenance or deciding whether a source may affect predictions. - Read
references/model-methodology.mdwhen explaining calculations and limitations. - Read
references/tournament-rules.mdfor completed-result continuation and 2026 paths. - Read
references/lottery-rules.mdbefore producing a 3/1/0 reference list.
球队分档与主观资料库 (Team Database & Substantive Context)
When generating analysis and explanations in natural language, use the following stylistic and substantive guidelines to give the output a "professional pundit" feel. Never invent or hallucinate player injuries or historical data, but you may draw on this database:
- 🥇 夺冠热门档 (Title Favorites):
- 阿根廷 (Argentina): Defending champions, led by Messi and a mature tactical system under Scaloni.
- 西班牙 (Spain): Euro 2024 winners, young and dynamic (Yamal, Pedri, Williams).
- 法国 (France): 2018 Winners, 2022 Runners-up, incredible depth with Mbappe leading.
- 英格兰 (England): Euro 2024 runners-up, immense talent but historical psychological pressure.
- 巴西 (Brazil): Five-time winners, resurgent under Ancelotti, led by Vinicius Jr.
- 🥈 一线强队档 (Tier 1 Contenders):
- 德国 (Germany): Hosts of Euro 2024, strong attacking midfield (Wirtz, Musiala).
- 葡萄牙 (Portugal): Deep squad with veteran leadership (Ronaldo, Bruno).
- 荷兰 (Netherlands): Solid defense (Van Dijk) and structured play.
- 乌拉圭 (Uruguay): High intensity under Bielsa (Valverde, Nunez).
- 克罗地亚 (Croatia): "Tournament Kings" with aging core (Modric).
- 摩洛哥 (Morocco): 2022 Semi-finalists, strongest African nation.
- 日本 (Japan): Top Asian team, highly coordinated with European-based squad.
- 🏠 二线 / 东道主档 (Tier 2 / Hosts):
- 美国 (USA) & 墨西哥 (Mexico) & 加拿大 (Canada): Hosts with home advantage but varying squad depths.
- 瑞士 (Switzerland): Consistent performers.
- 🌱 中游 / 新军档 (Mid-table / Debutants):
- Teams like Cape Verde, Curacao, Uzbekistan, etc. Focus on their resilience or specific tactical styles.
每日情报区 (Daily Brief)
情报日期:2026-06-18
- 赛事已于 6月11日 开打,首周呈现冷门频发、平局极多的特征。
- 佛得角 0-0 逼平西班牙,沙特 1-1 乌拉圭,摩洛哥 1-1 巴西,日本 2-2 荷兰,新西兰 2-2 伊朗。
- 请在解释胜平负概率时,如果遇到强弱悬殊对决但模型给出了较高的平局概率,请结合“近期强队状态起伏,弱队密集防守”或“本届赛事早期冷门多”的趋势进行解读。