bot-showalter

Generate user documentation from a completed rdm roadmap

Implement the next phase of an rdm roadmap

Review implementation of an rdm phase or task

Create an rdm roadmap with phases for a topic

Work on rdm tasks

Run fantasy baseball projection commands — predict, evaluate, compare systems, look up player projections/valuations, manage cached datasets, evaluate keeper league trades and optimization, run draft boards and mock drafts, and use Yahoo Fantasy integration for rosters, draft tracking, and keeper cost derivation. Use this when the user asks to run projections, compare systems, evaluate accuracy, look up a player, check valuations, manage/rebuild cached datasets, evaluate keeper league decisions/trades/optimization, view draft boards, run mock drafts, sync Yahoo league data, or manage Yahoo-derived keeper costs.

Run an autonomous feature-engineering experiment loop on the breakout-bust classification model. Tests candidate features against P(breakout) and P(bust) probability targets using marginal-value with auto-logging. Use when the user asks to "experiment on breakout-bust", "improve breakout-bust", or "explore features for breakout".

Run an autonomous feature-engineering experiment loop on the playing-time regression model. Tests candidate features against PA (batters) and IP (pitchers) targets using marginal-value with auto-logging. Use when the user asks to "experiment on playing-time", "improve playing-time", or "explore features for playing time".

Run an autonomous feature-engineering experiment loop on the statcast-gbm model (default). Analyzes residuals, generates feature hypotheses, screens via correlation, tests with fast feedback tools, logs results to the experiment journal, and validates winners. Use when the user asks to "experiment on", "explore features for", or "improve" the statcast-gbm model, or says "experiment on batter/pitcher" without specifying a model. Do NOT use for breakout-bust or playing-time — those have dedicated skills.

Review recent changes against the architectural principles in docs/principles.md. Flags semantic violations that static analysis cannot catch. Use when you want to check a commit, diff, or set of changes for principle violations.