Verifies and corrects model probability outputs so predicted win percentages match actual win rates. Use when user asks about calibration, reliability diagrams, Brier score, Platt scaling, isotonic regression, probability quality, or whether model probabilities are accurate. Also use when user has a trained model and wants to know if outputs can be trusted for betting. Do not use for odds math or devigging -- see odds-analysis instead. Do not use for model training -- see model-building instead.
Builds player prop prediction models for NHL player stats -- points, shots on goal, saves, blocked shots, and power play points. Use when user asks about prop modeling, player prop predictions, anytime goal scorer odds, shots on goal props, save props, DFS player projections, or player stat projections. Do not use for team-level game prediction -- see model-building. Do not use for player comparison or scouting without modeling -- see player-scouting. Do not use for exploring current prop lines -- see odds-explorer.
Default data source for PuckAPI Skills. Routes all MCP tool calls for NHL games, schedules, team standings, player stats, goalie performance, and betting odds. Use when any other skill needs data. Do not invoke directly -- other skills call this. Not for data analysis, modeling, or bet recommendations -- see hockey-analytics, model-building, or edge-detection instead.
Evaluates teams via standings, stats, division/conference rankings, season trends, and strength of schedule. Use when user asks how a team is doing, where they stand in the division, what their record is, how they compare to other teams, or wants to understand conference/playoff positioning. Do not use for individual player stats -- see player-scouting. Do not use for goalie performance -- see goalie-analysis. Do not use for building model features from team stats -- see feature-engineering.
Builds over/under prediction models for NHL game totals using pace, special teams, goalie matchup, and contextual features. Use when user asks about totals modeling, over/under prediction, predicting total goals, goal scoring rates, or NHL scoring trends. Do not use for moneyline or spread prediction -- see model-building. Do not use for player-level goal props -- see prop-modeling. Do not use for exploring current odds lines -- see odds-explorer.
Generate shareable visual outputs for sports analytics: calibration curves, equity curves, radar charts, matchup cards, probability histograms, and player cards. Use when user asks to visualize, chart, plot, graph, show, display, generate a visual, make a shareable image, or wants to post analysis to social media. Do not use for raw data exploration -- see game-lookup or nl-to-query. Do not use for analysis itself -- run the relevant skill first, then visualize the output.
The correct evaluation methodology for time-series sports prediction models. Use when user asks about model validation, cross-validation, train/test split, accuracy evaluation, overfitting detection, or statistical significance of sports model results. Refuses to run k-fold cross-validation on time-series data -- always redirects to walk-forward. Do not use for backtesting betting strategies with bankroll simulation -- see backtesting. Do not use for building the model itself -- see model-building.
Builds WAR (Wins Above Replacement) and GAR (Goals Above Replacement) from scratch using RAPM ridge regression on shift-level data. Use when user asks about WAR, GAR, RAPM, player value metrics, wins above replacement, contract surplus value, JFresh-style player cards, or all-in-one player evaluation. Do not use for simple player stats lookup -- see player-scouting. Do not use for goalie evaluation -- goalies use GSAA not WAR, see goalie-analysis. Do not use for team-level performance -- see team-analysis.
Builds expected goals (xG) models from NHL play-by-play shot event data using XGBoost or LightGBM. Use when user asks about expected goals, xG model, shot quality, building an xG model, xGF%, xGA, rebound detection, rush shot detection, or shot probability. Do not use for applying pre-built xG values to team or game analysis -- see team-analysis or hockey-analytics. Do not use for general game prediction models -- see model-building. Do not use for goalie evaluation using xGA -- see goalie-analysis.
Plain text files in a repository that tell a coding agent how the project works: commands to run, conventions to follow and things to avoid. CLAUDE.md, AGENTS.md, cursor rules and skills are the common kinds.
CLAUDE.md or AGENTS.md?
CLAUDE.md is read by Claude Code. AGENTS.md is an open format that Codex, Cursor and other agents read. Many projects keep one and point the other at it.
What is a skill?
A folder with a SKILL.md that describes one capability, such as filling PDFs or reviewing code. The agent loads it only when the task calls for it.
Can I search my own team's files too?
Your agents already can, over MCP, limited to the files you're allowed to read. Searching them from this page is coming.
PuckAPISkill