Summary
The dependency whose source you do not have becomes readable in the session.
The three entry points match how bytecode actually turns up: a loose .class file, a JAR sitting in a local Maven repository, or a fully qualified package name you already have from a stack trace. There is no search step — you supply the location or the name, and the JAR route needs the JAR path and the class name together.
What it is
A Java decompiler behind three tools, addressed by file path, by JAR plus class name, or by package name.
What you get
- A .class file decompiled from its path on disk
- One class decompiled out of a JAR, by naming the JAR file and the class inside it
- Any class resolved by package name against a classpath passed per call or the CLASSPATH environment variable
Requirements
The bytecode itself — a .class file, a JAR, or a package resolvable on a classpath — and CLASSPATH set if you would rather not pass one on every call.
Setup effort
One command — npx -y @idachev/mcp-javadc