java-chunker

What problem does it solve?

Large Java source files often exceed LLM context limits and lose important class or method-level context when naively truncated. This Skill splits Java files at semantic boundaries and preserves class headers so each piece remains self-contained and meaningful for downstream migration, validation, test generation, and documentation agents.

Core Features & Use Cases

• Semantic Chunking: Split at class and method boundaries, never inside a method body, and prepend the class header (package, imports, class signature) to every chunk so chunks are context-aware.
• Structured Output: Return per-chunk metadata (chunk index, total chunks, class name, file path) so agents can track piecewise translations and reassemble results.
• Prompt & File Utilities: Load and validate prompt templates with required variables, convert Java paths and CamelCase names to Python module paths, parse Maven/Gradle dependencies, and read/write project files safely.
• Use Cases: Migrate very large Java classes to idiomatic Python without losing semantics, enable per-chunk validation and test generation, and support prompt building for multilingual agents.

Quick Start

Chunk the large Java source file into context-preserving pieces that include the class header and fit within a 3000-token model context.