review-delta

Skill

A PR review skill powered by code-review-graph, which builds a structural map of your codebase using Tree-sitter and provides precise context to AI assistants via MCP, achieving an average 8.2x token reduction compared to full-codebase scanning.

A skill for navigating and understanding codebase structure using code-review-graph MCP tools, including architecture overviews, community detection, semantic search, and relationship tracing (callers, callees, imports). Designed for token-efficient exploration in five tool calls or fewer.

Performs structured code reviews using change detection, impact analysis, and a knowledge graph for risk-scored analysis and blast radius assessment.

Code-review-graph uses Tree-sitter to build a structural code map and tracks changes incrementally for precise AI context, achieving 8.2x average token reduction.

A code review tool that builds a structural map of your codebase using Tree-sitter, so AI coding tools only read what they need instead of re-reading everything on every task. Reduces token usage by up to 70% by providing dependency graphs, function signatures, and change-scoped context for more efficient AI-powered code reviews.