finding-duplicate-functions

What problem does it solve? LLM-generated and organically grown codebases accumulate semantic duplicates: functions that do the same thing under different names or implementations. Classical copy-paste detectors like jscpd only find syntactic duplicates and miss same-intent code, leading to bloated utilities and inconsistent behavior. ## Core Features & Use Cases - Two-Phase Detection Pipeline: Extracts a function catalog with ripgrep and jq, then uses LLM subagents to categorize functions by domain and detect duplicates within each category. - Model-Tiered Analysis: Uses a lightweight model (haiku-class) for cheap categorization and a stronger model (opus-class) for subtle semantic duplicate detection, with confidence-ranked reporting. - Prioritized Markdown Report: Aggregates per-category findings into a report grouped by confidence level so reviewers know which consolidations to act on first. - Use Case: Before a major refactor of an LLM-generated TypeScript app, run the pipeline over app/components/ and app/lib/ to find reimplemented validation, formatting, and path utilities, then consolidate them safely behind tests. ## Quick Start Ask the assistant to audit the src directory for semantically duplicate functions and produce a prioritized consolidation report.