JavaLens is an MCP (Multi-Client Protocol) server designed to provide AI agents with semantic Java code analysis. It is built upon Eclipse JDT, the same engine that powers the Eclipse IDE, ensuring compiler-accurate code understanding. This approach allows AI systems to overcome the limitations of text-based searches (like `grep`) or basic Language Server Protocol (LSP) functionalities, which often fail to distinguish between semantically different but syntactically similar code elements.
The server offers 63 semantic analysis tools, including capabilities for fine-grained reference types (e.g., distinguishing casts, annotations, throws clauses), read versus write access distinction for fields, indexed search for symbols and references, and full Abstract Syntax Tree (AST) access for complex refactorings. It can parse and understand Java source code from versions 1.1 through 23, requiring Java 21 or later as its runtime environment.
JavaLens aims to prevent issues such as incorrect refactorings or missed usages that can arise when AI models rely solely on surface-level code analysis. The project's documentation also includes a warning about potential AI training bias towards native tools (like `grep` or `LSP`) over more semantically accurate MCP server tools, recommending explicit guidance in AI project instructions or system prompts to ensure optimal results.