Code Superman
A codebase-understanding tool for AI agents, distributed as an MCP server and command-line tool, that turns a repository into structured static-analysis data, such as a tech-stack summary, symbol outlines and a dependency graph, for a language model to interpret itself.
Code Superman performs offline static analysis rather than relying on a hosted API or requiring an API key, since the caller is expected to be a large language model itself. Its three MCP tools report language distribution, detected technology stack, entry points, a token-count estimate, complexity metrics and core modules for a project; return a tree-sitter-based symbol outline of functions and classes for a single file; and export the whole analysis as an architecture mind map covering an overview, entry points, core modules, warnings and the directory tree.
Static analysis parses common manifest files, such as package.json, Cargo.toml, go.mod, requirements.txt, pom.xml and .csproj, against more than 100 built-in framework-detection rules, and tree-sitter parsing covers twelve languages including Rust, Python, TypeScript, JavaScript, Go, Java, C, C++, C# and PHP. An analysis-detail flag controls how much is returned, from a brief summary up to full dependency edges and metrics.
It is written in Rust and released under the Apache-2.0 license.