Pr-surgeon

PR Surgeon: Intelligent Pull Request Decomposition In enterprise software development—especially during legacy migrations or large refactors—engineering teams often face the critical bottleneck of "monster" Pull Requests. A single PR containing 300+ modified files can stall development for weeks, as no human reviewer can safely hold the full diff and its cascading side-effects in their head. PR Surgeon is an intelligent developer tool built to solve this exact problem by automatically decomposing massive PRs into a series of small, safe, and highly reviewable sub-PRs. Unlike naive chunking tools that arbitrarily split files, PR Surgeon actually understands your code's architecture. It parses the Abstract Syntax Trees (AST) of the modified files and constructs a directed dependency graph representing true import and reference relationships. Using community detection algorithms, it identifies tightly-coupled clusters of files and calculates a topological sort to determine the safest merging order (e.g., database schemas first, API logic next, frontend UI last). To finalize the pipeline, PR Surgeon leverages IBM watsonx.ai Granite models to semantically enrich each proposed sub-PR. The AI generates context-aware, human-readable descriptions, performs risk assessments, and recommends ideal reviewers. Through an interactive React Flow dashboard, developers can visualize the dependency graph and export the final decomposition plan, transforming an unreviewable monolith into an organized, trackable pipeline of iterative updates.