What problem does it solve? During AI-assisted coding sessions, work often drifts beyond the original plan, introducing unplanned features, unrelated refactors, and extra dependencies. This Skill quantifies that drift so orchestrators and users can catch scope creep before it compounds. ## Core Features & Use Cases - Plan vs. Actual Comparison: Loads the planned scope from TodoWrite tasks or .rune/progress.md and compares it against git diff output (staged and unstaged). - Drift Quantification: Computes a drift percentage and classifies it into four tiers: ON_TRACK, MINOR_DRIFT, SIGNIFICANT_DRIFT, and OUT_OF_CONTROL, each with a defined action. - Smart Classification Rules: Treats test files, config side-effects, and lock files as in-scope natural dependencies while flagging unplanned features and unrelated refactors. - Use Case: An L1 orchestrator finishes a parallel workstream and invokes this Skill before merging; it reports 40% drift caused by an unplanned component and recommends reverting two files before continuing. ## Quick Start Ask the agent to run a scope check comparing current git changes against the original plan and report the drift percentage with recommendations.
