What problem does it solve? Repeated plan rejections contain valuable feedback, but that signal is buried across dozens or hundreds of archived files. This Skill reads every denied plan, clusters the reviewer's feedback into a taxonomy, and turns it into concrete prompt instructions that reduce future denials. ## Core Features & Use Cases - Map-reduce pattern analysis: Parallel extraction agents read every denied plan file, then reduction agents build a denial taxonomy, recurring phrase counts, and evolution-over-time analysis. - HTML dashboard report: Produces a self-contained, versioned report with narrative findings, KPIs, taxonomy bars, and a copyable corrective prompt derived from real denial data. - Dual data sources: Works with Plannotator plan archives first, falling back to Claude Code ExitPlanMode denial reasons extracted from JSONL transcripts via a bundled parser. - Use Case: After months of rejecting agent plans, run this Skill to discover that 25% of your denials are for missing narrative overviews, then install an improvement hook that injects corrective instructions into every future planning session. ## Quick Start Analyze my Plannotator plan archive and generate a compound planning report of my denial patterns.
